issue_owner_repo listlengths 2 2 | issue_body stringlengths 0 261k ⌀ | issue_title stringlengths 1 925 | issue_comments_url stringlengths 56 81 | issue_comments_count int64 0 2.5k | issue_created_at stringlengths 20 20 | issue_updated_at stringlengths 20 20 | issue_html_url stringlengths 37 62 | issue_github_id int64 387k 2.46B | issue_number int64 1 127k |
|---|---|---|---|---|---|---|---|---|---|
[
"MonetDB",
"MonetDB"
] | Date: 2006-04-08 10:35:55 +0200
From: @drstmane
To: @sjoerdmullender
Version: -- development
Last updated: 2006-11-02 08:31:58 +0100
## Comment 5222
Date: 2006-04-08 10:35:55 +0200
From: @drstmane
/bin/bash ../../../libtool --mode=install
/usr/bin/install -c 'lib_pcre.la'
'/var/tmp/cwi.GNU.32.32.d.22509/MonetDB/.GNU.32.32.d-CYGWIN1.5.12/lib/MonetDB/lib_pcre.la'
libtool: install: warning: relinking `lib_pcre.la'
(cd
/var/tmp/cwi.GNU.32.32.d.22509/MonetDB/CYGWIN1.5.12/src/modules/plain;
/bin/bash ../../../libtool --tag=CC --mode=relink gcc
-O2 -std=gnu99 -Wall -W -O6 -finline-functions
-falign-loops=4 -falign-jumps=4 -falign-functions=4
-fexpensive-optimizations -funroll-loops
-frerun-cse-after-loop -frerun-loop-opt
-Werror-implicit-function-declaration -Werror
-Wno-unused-function -Wno-unused-label
-Wno-strict-aliasing -D_REENTRANT -no-undefined -o
lib_pcre.la -rpath
/var/tmp/cwi.GNU.32.32.d.22509/MonetDB/.GNU.32.32.d-CYGWIN1.5.12/lib/MonetDB
lib_pcre_la-pcre.glue.lo lib_pcre_la-pcre.lo
../../common/libstream.la ../../gdk/libbat.la
../../monet/libmonet.la -lws2_32 -lz -lbz2 -lpthread
-L/usr/lib -lpcre )
gcc -shared .libs/lib_pcre_la-pcre.glue.o
.libs/lib_pcre_la-pcre.o
-L/var/tmp/cwi.GNU.32.32.d.22509/MonetDB/CYGWIN1.5.12/src/gdk/.libs
-L/var/tmp/cwi.GNU.32.32.d.22509/MonetDB/CYGWIN1.5.12/src/common/.libs
-L/var/tmp/cwi.GNU.32.32.d.22509/MonetDB/.GNU.32.32.d-CYGWIN1.5.12/lib
-lstream -lbat -lmonet -lws2_32 -lz -lbz2 -lpthread
-L/usr/lib -lpcre -o .libs/cyg_pcre-0.dll
-Wl,--image-base=0x10000000
-Wl,--out-implib,.libs/lib_pcre.dll.a
Creating library file: .libs/lib_pcre.dll.a
Info: resolving _chr_nil by linking to __imp__chr_nil
(auto-import)
Info: resolving _str_nil by linking to __imp__str_nil
(auto-import)
Info: resolving _BATatoms by linking to __imp__BATatoms
(auto-import)
Info: resolving _pcre_malloc by linking to
__imp__pcre_malloc (auto-import)
Info: resolving _pcre_free by linking to
__imp__pcre_free (auto-import)
/usr/bin/install -c .libs/lib_pcre.dll.aT
/var/tmp/cwi.GNU.32.32.d.22509/MonetDB/.GNU.32.32.d-CYGWIN1.5.12/lib/MonetDB/lib_pcre.dll.a
base_file=`basename ${file}`
dlpath=`/bin/bash 2>&1 -c '.
.libs/'${base_file}'i;echo $dlname'`
dldir=/var/tmp/cwi.GNU.32.32.d.22509/MonetDB/.GNU.32.32.d-CYGWIN1.5.12/lib/MonetDB/`dirname
$dlpath`
dirname: too many arguments
Try `dirname --help' for more information.
make[8]: *** [install-pcreLTLIBRARIES] Error 1
See also
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.GNU.32.32.d-CYGWIN1.5.12/install.out
## Comment 5223
Date: 2006-06-19 14:40:34 +0200
From: @sjoerdmullender
Logged In: YES
user_id=43607
Is this still a problem? I have many problems compiling on
Cygwin, but not this one.
## Comment 5224
Date: 2006-11-02 20:31:58 +0100
From: @drstmane
Logged In: YES
user_id=572415
Clean-up: I don't do any CYGWIN compilation/testing any
more, hence, this is no problem any more.
## Comment 5225
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1466773 at http://sourceforge.net/support/tracker.php?aid=1466773
| MonetDB's `make install` fails on CYGWIN | https://api.github.com/repos/MonetDB/MonetDB/issues/910/comments | 0 | 2020-11-30T09:05:37Z | 2024-06-28T13:33:33Z | https://github.com/MonetDB/MonetDB/issues/910 | 753,293,721 | 910 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-04-08 12:38:18 +0200
From: @drstmane
To: Jan Rittinger <<jan.rittinger>>
Version: -- development
Last updated: 2006-06-20 09:36:14 +0200
## Comment 5213
Date: 2006-04-08 00:38:18 +0200
From: @drstmane
test tests/XQuery/Tests/bool7.xq fails after recent
changes; see
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XQuery/bool7.out.00.html
for details.
## Comment 5214
Date: 2006-05-02 13:07:47 +0200
From: @drstmane
Logged In: YES
user_id=572415
Samuel,
any news on this one?
Stefan
## Comment 5215
Date: 2006-05-02 14:19:43 +0200
From: @drstmane
Logged In: YES
user_id=572415
As Jens just pointed out, Samuel indeed just extended the
test and the stable output accordingly:
========
Index: tests/XQuery/Tests/bool7.xq
===================================================================
RCS file:
/cvsroot/monetdb/pathfinder/tests/XQuery/Tests/bool7.xq,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -r1.1 -r1.2
--- tests/XQuery/Tests/bool7.xq 20 Jan 2005 15:08:51 -0000
1.1
+++ tests/XQuery/Tests/bool7.xq 6 Apr 2006 13:49:19 -0000
1.2
@@ -1 +1 @@
-for $b in (1,0) return if ($b) then for $a in
(0,1,2,"foo","",0.0,1.3,boolean(1),boolean(0)) return
boolean($a) else boolean (())
+for $b in (1,0) return if ($b) then for $a in
(0,1,2,"foo","",0.0,1.3,boolean(1),boolean(0),<a/>) return
boolean($a) else boolean (())
========
Index: tests/XQuery/Tests/bool7.stable.out
===================================================================
RCS file:
/cvsroot/monetdb/pathfinder/tests/XQuery/Tests/bool7.stable.out,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -r1.2 -r1.3
--- tests/XQuery/Tests/bool7.stable.out 12 May 2005 17:53:08
-0000 1.2
+++ tests/XQuery/Tests/bool7.stable.out 6 Apr 2006 13:49:18
-0000 1.3
@@ -23,7 +23,7 @@
<?xml version="1.0" encoding="utf-8"?>
<XQueryResult>
-false true true true false false true true false false
+false true true true false false true true false true false
</XQueryResult>
========
Both look fine to me.
My bug report was triggered by the fact that this test fails
since April 07 2006, i.e., related to changes made on April
06 2006 --- one of these changes was Samuels test extension,
but apparely, this does not cause the test to fail, but was
just done by coincidence on the same day that the test
started failing.
Hence, either Samuel's changes to the test happen to trigger
an older bug that was no triggered by the original test, or
some of the other changes of April 06 2006 introduced a bug
that causes this test to fail.
On April 06 2006, JanF, JanR, SamuelK amd PeterB checked in
changes (see details below); hence, each of them should
check whether his changes might have caused test
tests/XQuery/Tests/bool7.xq to fail (see
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XQuery/bool7.out.00.html
for details).
Let's start with JanR ...
===================================================================
2006/04/06 - jflokstra: compiler/mil/milprint_summer.c,1.190
compiler/semantics/xquery_fo.c,1.60
modules/pftijah/Makefile.ag,1.3
modules/pftijah/nexi.c,1.1 modules/pftijah/nexi.h,1.1
modules/pftijah/nexi.l,1.1 modules/pftijah/nexi.y,1.1
modules/pftijah/nexi_generate_mil.c,1.1
modules/pftijah/nexi_generate_mil.h,1.1
modules/pftijah/nexi_generate_plan.c,1.1
modules/pftijah/nexi_generate_plan.h,1.1
modules/pftijah/nexi_preprocessor.c,1.1
modules/pftijah/nexi_rewriter.c,1.1
modules/pftijah/nexi_rewriter.h,1.1
modules/pftijah/pftijah.mx,1.3
modules/pftijah/pftijah_tokenize.l,1.2
modules/pftijah/serialize_pftijah.mx,1.2
runtime/serialize.mx,1.56
- add out of bounds check to serialize.mx WS BAT lookup.
There is still a bug
in print_doc() but this one now terminates with an error
message instead of a
crash.
- add 2 new XQUERY functions to xquery_fo and
milprint_summer. They are
guarded by ifdef HAVE_PFTIJAH so they should not be
visible in
regular Pathfinder systems. The functions are:
fn:tijah-command(string?) as boolean
fn:tijah-query(item*, string) as item*
- integrated all of Vojkans Tijah parser stuff in pftijah.
When the print_doc()
bug is solved we are ready to run our first real Tijah
query on Pathfinder.
Jan F.
===================================================================
2006/04/06 - jflokstra: runtime/serialize.mx,1.57
* try to repair the print_doc() bug.
The ws_documentIndex() now uses the extra oid indirection
through CONT_DOCID.
It now works for my test scripts but I am not really sure
if it works OK
for all cases because I do not use the new CONT_COLL yet.
I will contact
Peter B. about this
===================================================================
2006/04/06 - tsheyar: compiler/algebra/opt/opt_icol.c,1.4
-- fixed bug (empty tables have to be treated different than
literal tables)
Thanks to Samuel!
===================================================================
2006/04/06 - kerschbaumers: compiler/Makefile.ag,1.22
compiler/compile.c,1.71
compiler/main.c,1.56 compiler/algebra/algebra.c,1.40
compiler/algebra/algebra_cse.c,1.32
compiler/algebra/builtins.c,1.34
compiler/algebra/core2alg.brg,1.24
compiler/algebra/logical.c,1.17
compiler/algebra/planner.c,1.18
compiler/debug/logdebug.c,1.22
compiler/debug/physdebug.c,1.19
compiler/include/Makefile.ag,1.24
compiler/include/algebra.h,1.37
compiler/include/bitset.h,1.1
compiler/include/logical.h,1.14
compiler/include/mil.h,1.26
compiler/include/mil_dce.h,1.1
compiler/include/pathfinder.h,1.38
compiler/mem/Makefile.ag,1.12 compiler/mem/array.c,1.17
compiler/mem/bitset.c,1.1 compiler/mil/Makefile.ag,1.19
compiler/mil/mil.c,1.23 compiler/mil/mil_dce.c,1.1
compiler/mil/milgen.brg,1.29
compiler/mil/milprint.c,1.33
compiler/semantics/xquery_fo.c,1.61
- dead code elimination added
- changed mil identifiers from strings to ints
- implementation for a dynamic bitset added
- algebraic type aat_uA added (atomicUntyped)
- changed physdebug.c to use the same colors, identifiers as
logdebug.c
===================================================================
2006/04/06 - kerschbaumers:
tests/XQuery/Tests/bool7.stable.out,1.3
tests/XQuery/Tests/bool7.xq,1.2
- added a node to the boolean() test function
===================================================================
2006/04/06 - kerschbaumers: compiler/debug/physdebug.c,1.20
- changed physdebug.c to look more like logdebug.c
===================================================================
2006/04/06 - tsheyar: compiler/semantics/xquery_fo.c,1.62
-- make functions fn:zero-or-one and fn:exactly-one
conformant to the
recommendation (type: item* instead of node*).
-- remove unused function
===================================================================
2006/04/06 - tsheyar: compiler/core/coreopt.brg,1.27
-- typed-value should cast to untypedAtomic instead of
introducing only
a seqcast. (Since Samuel's checkin untypedAtomic can also
be evaluated
in the algebra version :-).)
===================================================================
2006/04/06 - boncz: compiler/mil/milprint_summer.c,1.191
runtime/pathfinder.mx,1.185
- fixed issues when multiple documents would be loaded in
the working set
sjoerd: this may affect your Update tessts that have hard
MIL in them
===================================================================
## Comment 5216
Date: 2006-05-02 15:58:54 +0200
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
In function fn_boolean() only atomic types are handled.
Therefore fn:boolean on nodes will _always_ return the wrong
result. If someone is willing to extend this method I have
no objections.
## Comment 5217
Date: 2006-05-02 16:08:11 +0200
From: @drstmane
Logged In: YES
user_id=572415
IMHO, "even" atomic types are handled incorrectly, now,
i.e., since April 06 2006:
the query
for $b in (1,0) return if ($b) then for $a in
(0,1,2,"foo","",0.0,1.3,boolean(1),boolean(0)) return
boolean($a) else boolean (())
used to (IMHO correctly) return
false true true true false false true true false false
before April 06 2006; after the changes of April 06 2006
(both to the test and various code)
the query
for $b in (1,0) return if ($b) then for $a in
(0,1,2,"foo","",0.0,1.3,boolean(1),boolean(0),<a/>) return
boolean($a) else boolean (())
(just the node <a/> has been added to the sequence)
now (IMHO incorrectly) return
true true true true true true true true true true false
("true" in all cases except the empty secquence ())
... or am I wrong, here ... ?
## Comment 5218
Date: 2006-05-02 16:12:42 +0200
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
In function fn_boolean() only atomic types are handled.
Therefore fn:boolean on nodes will _always_ return the wrong
result. If someone is willing to extend this method I have
no objections.
## Comment 5219
Date: 2006-05-02 16:20:22 +0200
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
oops, shouldn't have reused reload - the result stays the
same :)
The test only fails if nodes and atomic values are mixed
(test is based on PFty_subtype which fails for this mix). As
this is a very unlikely scenario which I don't think is
important to support with milprint_summer the only solution
for a green testweb entry is a modified test (or it stays
red until the algebra version becomes default :)).
## Comment 5220
Date: 2006-06-20 09:36:13 +0200
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
I thought this pending bug reports are closed automatically
after some time...
## Comment 5221
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1466635 at http://sourceforge.net/support/tracker.php?aid=1466635
| PF: test tests/XQuery/Tests/bool7.xq fails | https://api.github.com/repos/MonetDB/MonetDB/issues/909/comments | 0 | 2020-11-30T09:05:33Z | 2024-06-28T13:33:33Z | https://github.com/MonetDB/MonetDB/issues/909 | 753,293,676 | 909 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-04-07 11:54:21 +0200
From: @drstmane
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-05-27 11:38:58 +0200
## Comment 5209
Date: 2006-04-07 23:54:21 +0200
From: @drstmane
The tests of swizzle() & remap() in
src/modules/plain/Tests/bat.milM &
src/modules/plain/Tests/bat2.milM
need to be checked.
[swizzle]() triggers BATproperty errors:
!WARNING: BATpropcheck: BAT tmp_52 was incorrectly
marked sorted!
!WARNING: BATpropcheck: BAT tmp_52 remains marked
radix-clustered on 32 bits; not checked!
The results od remap() differ from the original one;
either the new ones (and hence the code) are wrong and
need to be fixed,
or the old ones are obsolete, and hence, the new ones
need to be approved as stable.
See
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/src_modules_plain/bat.out.00.html
and
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/src_modules_plain/bat2.out.00.html
for details.
## Comment 5210
Date: 2006-05-02 13:08:14 +0200
From: @drstmane
Logged In: YES
user_id=572415
Peter,
any news on this one?
Stefan
## Comment 5211
Date: 2006-05-27 23:38:57 +0200
From: @drstmane
Logged In: YES
user_id=572415
Fixed by Peter on 2006/05/23.
## Comment 5212
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1466609 at http://sourceforge.net/support/tracker.php?aid=1466609
| tests of swizzle() & remap() seem to fail | https://api.github.com/repos/MonetDB/MonetDB/issues/908/comments | 0 | 2020-11-30T09:05:30Z | 2024-06-27T11:09:49Z | https://github.com/MonetDB/MonetDB/issues/908 | 753,293,635 | 908 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-04-07 10:50:29 +0200
From: @drstmane
To: Jan Rittinger <<jan.rittinger>>
Version: -- development
Last updated: 2006-04-09 12:23:56 +0200
## Comment 5205
Date: 2006-04-07 22:50:29 +0200
From: @drstmane
gcc -m64 -DHAVE_CONFIG_H -I.
-I/ufs/manegold/_/scratch0/Monet/Testing/Current/pathfinder/compiler/algebra
-I../..
-I/ufs/manegold/_/scratch0/Monet/Testing/Current/pathfinder/compiler/algebra
-I../include
-I/ufs/manegold/_/scratch0/Monet/Testing/Current/pathfinder/compiler/algebra/../include
-I/ufs/manegold/_/scratch0/Monet/Testing/Current/.GNU.64.64.d..PREFIX./MONETDB/include/MonetDB
-I/ufs/manegold/_/scratch0/Monet/Testing/Current/.GNU.64.64.d..PREFIX./MONETDB/include/MonetDB/common
-I/ufs/manegold/_/scratch0/Monet/Testing/Current/.GNU.64.64.d..PREFIX./MONETDB/include/MonetDB/gdk
-I/ufs/manegold/_/scratch0/Monet/Testing/Current/.GNU.64.64.d..PREFIX./MONETDB/include/MonetDB/monet
-I/ufs/manegold/_/scratch0/Monet/Testing/Current/.GNU.64.64.d..PREFIX./MONETDB/include/MonetDB/plain
-I/ufs/manegold/_/scratch0/Monet/Testing/Current/.GNU.64.64.d..PREFIX./MONETDB/include/MonetDB/contrib
-I/ufs/manegold/_/scratch0/Monet/Testing/Current/.GNU.64.64.d..PREFIX./MONETDB/include/MonetDB/mapi
-I/ufs/manegold/_/scratch0/Monet/Testing/Current/.GNU.64.64.d..PREFIX./MONETDB/include/MonetDB/C
-DLIBALGEBRA -g -O2 -std=c99 -Wall -W
-Werror-implicit-function-declaration -Werror
-Wno-unused-function -Wno-unused-label -c
/ufs/manegold/_/scratch0/Monet/Testing/Current/pathfinder/compiler/algebra/properties.c
-fPIC -DPIC -o .libs/libalgebra_la-properties.o
/ufs/manegold/_/scratch0/Monet/Testing/Current/pathfinder/compiler/algebra/properties.c:
In function 'PFprop':
/ufs/manegold/_/scratch0/Monet/Testing/Current/pathfinder/compiler/algebra/properties.c:82:
error: 'struct PFprop_t' has no member named 'unq_names'
make[6]: *** [libalgebra_la-properties.lo] Error 1
## Comment 5206
Date: 2006-04-07 23:06:57 +0200
From: @drstmane
Logged In: YES
user_id=572415
The below patch (checked in to cvs) seems to make it compile.
Jan,
could you please check, whether this is also the correct fix?
========
Index: compiler/algebra/properties.c
===================================================================
RCS file:
/cvsroot/monetdb/pathfinder/compiler/algebra/properties.c,v
retrieving revision 1.17
diff -u -r1.17 properties.c
--- compiler/algebra/properties.c 7 Apr 2006 15:53:58
-0000 1.17
+++ compiler/algebra/properties.c 7 Apr 2006 21:03:24
-0000
@@ -79,7 +79,7 @@
ret->r_domains = PFarray (sizeof (dom_pair_t));
/* initialize unique name information */
- ret->unq_names = PFarray (sizeof (name_pair_t));
+ ret->name_pairs = PFarray (sizeof (name_pair_t));
return ret;
}
========
## Comment 5207
Date: 2006-04-09 12:23:56 +0200
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
seems to be fine - thx Stefan!
## Comment 5208
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1466572 at http://sourceforge.net/support/tracker.php?aid=1466572
| PF: compiler/algebra/properties.c fails to compile | https://api.github.com/repos/MonetDB/MonetDB/issues/907/comments | 0 | 2020-11-30T09:05:27Z | 2024-06-27T11:09:48Z | https://github.com/MonetDB/MonetDB/issues/907 | 753,293,584 | 907 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-04-05 04:24:13 +0200
From: @drstmane
To: @sjoerdmullender
Version: -- development
Last updated: 2006-04-07 04:32:28 +0200
## Comment 5202
Date: 2006-04-05 04:24:13 +0200
From: @drstmane
icc -no-gcc -DHAVE_CONFIG_H -I.
-I/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet
-I../..
-I/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet
-I../common
-I/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet/../common
-I../gdk
-I/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet/../gdk
-DLIBMONET -we140 -wd1418 -c99 -Wall -w2 -mp1 -O3
-restrict -unroll -tpp7 -axWP -we266 -Werror
-wd1418,1419,279,310,981,810,444,193,111,177,171,181,764,269,108,188,1357,102,70,1572
-D_REENTRANT -c monet.c -KPIC -DPIC -o
.libs/libmonet_la-monet.o
/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet/monet.mx(423):
error 187: use of "=" where "==" may have been intended
VARfreeze(v);
^
/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet/monet.mx(423):
error 187: use of "=" where "==" may have been intended
VARfreeze(v);
^
/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet/monet.mx(423):
error 187: use of "=" where "==" may have been intended
VARfreeze(v);
^
/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet/monet.mx(423):
error 187: use of "=" where "==" may have been intended
VARfreeze(v);
^
/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet/monet.mx(423):
error 187: use of "=" where "==" may have been intended
VARfreeze(v);
^
/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet/monet.mx(423):
error 187: use of "=" where "==" may have been intended
VARfreeze(v);
^
compilation aborted for monet.c (code 2)
make[5]: *** [libmonet_la-monet.lo] Error 1
icc -no-gcc -DHAVE_CONFIG_H -I.
-I/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet
-I../..
-I/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet
-I../common
-I/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet/../common
-I../gdk
-I/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet/../gdk
-DLIBMONET -we140 -wd1418 -c99 -Wall -w2 -mp1 -O3
-restrict -unroll -tpp7 -axWP -we266 -Werror
-wd1418,1419,279,310,981,810,444,193,111,177,171,181,764,269,108,188,1357,102,70,1572
-D_REENTRANT -c monet_tbl.c -KPIC -DPIC -o
.libs/libmonet_la-monet_tbl.o
/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/monet/monet_tbl.mx(1423):
error 187: use of "=" where "==" may have been intended
(void) VARfreeze(v);
^
compilation aborted for monet_tbl.c (code 2)
make[5]: *** [libmonet_la-monet_tbl.lo] Error 1
icc -no-gcc -DHAVE_CONFIG_H -I.
-I/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/modules/plain
-I../../..
-I/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/modules/plain
-I../../common
-I/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/modules/plain/../../common
-I../../gdk
-I/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/modules/plain/../../gdk
-I../../monet
-I/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/modules/plain/../../monet
-DLIBBUILTIN -we140 -wd1418 -c99 -Wall -w2 -mp1 -O3
-restrict -unroll -tpp7 -axWP -we266 -Werror
-wd1418,1419,279,310,981,810,444,193,111,177,171,181,764,269,108,188,1357,102,70,1572
-D_REENTRANT -c builtin.c -KPIC -DPIC -o
.libs/lib_builtin_la-builtin.o
/export/scratch0/monet/monet.ntv.64.64.d.19112/MonetDB/src/modules/plain/builtin.mx(626):
error 187: use of "=" where "==" may have been intended
VARfreeze(act);
^
compilation aborted for builtin.c (code 2)
make[6]: *** [lib_builtin_la-builtin.lo] Error 1
see also
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.ntv.64.64.d-Fedora4/make.out
## Comment 5203
Date: 2006-04-07 16:32:28 +0200
From: @njnes
Logged In: YES
user_id=43556
Last nights testing was succesfull again
## Comment 5204
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1464628 at http://sourceforge.net/support/tracker.php?aid=1464628
| MonetDB fails to compile with icc | https://api.github.com/repos/MonetDB/MonetDB/issues/906/comments | 0 | 2020-11-30T09:05:24Z | 2024-06-27T11:09:47Z | https://github.com/MonetDB/MonetDB/issues/906 | 753,293,548 | 906 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-04-03 04:02:06 +0200
From: Joris van Rantwijk <<jvrantwijk>>
To: GDK devs <<bugs-common>>
Version: -- development
Last updated: 2007-02-19 04:37:04 +0100
## Comment 5196
Date: 2006-04-03 16:02:06 +0200
From: Joris van Rantwijk <<jvrantwijk>>
It seems that I can not pass a nodelist to a user
defined function if this nodelist is the result of an
XPath expression:
I tried the following query:
declare namespace joris='http://joris.net/';
declare function joris:f($n as node()*) as node()* {
$n
};
joris:f( <doc><p n="aap"/><p n="noot"/><p
n="mies"/></doc> // p )
and this returns:
<p n="aap"/>
while i expected it to return:
<p n="aap"/> ,
<p n="noot"/> ,
<p n="mies"/>
and in fact it does return the expected result when run
on the Saxon engine.
The following query however:
declare namespace joris='http://joris.net/';
declare function joris:f($n as node()*) as node()* {
$n
};
joris:f( ( <aap/> , <noot/> , <mies/> ) )
returns:
<aap/>,
<noot/>,
<mies/>
as expected.
So it seems that a nodelist passed as a function
argument is sometimes, but not always, silently
truncated to its first element.
Tested on Pathfinder CVS head, compiled on Linux i386,
gcc 3.4.4, autoconf 2.59, automake 1.9.2, libtool 1.5.6.
## Comment 5197
Date: 2006-04-07 11:31:23 +0200
From: @Teggy
Logged In: YES
user_id=642632
Hi all,
an observation about this bug. Evaluating
declare function local:foo($n as node()*) as node()*
{
$n
};
let $t := <a><b/><c/><d/></a>
return
local:foo($t/descendant-or-self::node()/child::*)
exhibits the bug mentioned by Joris: the function returns
the first node </b> only. Evaluating
declare function local:foo($n as node()*) as node()*
{
$n
};
let $t := <a><b/><c/><d/></a>
return
local:foo($t/child::*)
(which is equivalent in this particular case) returns the
correct result: <b/><c/></d>.
Regards,
--Teggy
## Comment 5198
Date: 2006-04-07 12:01:12 +0200
From: @Teggy
Logged In: YES
user_id=642632
Materialization-related bug in UDF functional call preparation.
(Actual bug fix by Jan R.)
## Comment 5199
Date: 2006-11-29 10:59:06 +0100
From: @drstmane
Logged In: YES
user_id=572415
Originator: NO
Re-opened, as no test has been provided, yet.
User (submitter) and/or developer (assignee),
please supply a respective test,
or explain why a test is not required/possible.
## Comment 5200
Date: 2007-02-19 16:37:04 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Originator: NO
Test added.
## Comment 5201
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1463558 at http://sourceforge.net/support/tracker.php?aid=1463558
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-common@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
| wrong result when passing nodelist to function | https://api.github.com/repos/MonetDB/MonetDB/issues/905/comments | 0 | 2020-11-30T09:05:21Z | 2024-06-27T11:09:46Z | https://github.com/MonetDB/MonetDB/issues/905 | 753,293,508 | 905 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-28 02:32:06 +0200
From: Wouter Alink <<alink>>
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-06-19 10:10:05 +0200
## Comment 5190
Date: 2006-03-28 14:32:06 +0200
From: Wouter Alink <<alink>>
probably related to: [1339844] "a parent step into the
void"
I have not been following all bugs lately, so I am not
sure whether this bug still exists (I'm using 0.10.3
from a few weeks ago).
given the query (which i don't know whether it is valid):
for $i in doc("test.xml")//b/@c
where $i = "d"
return $i/parent::*
and the document:
<a>
<b c="e"/>
<a>
<a>
<a>
<b c="d"/>
</a>
<b c="d"/>
</a>
</a>
</a>
returns the whole document.
If the document is changed to:
<a>
<a>
<a>
<a>
<b c="d"/>
</a>
<b c="d"/>
</a>
</a>
</a>
the query returns the empty result.
And if the document modified to:
<a>
<b c="e"/>
<a>
<a>
<a>
<b c="d"/>
</a>
<b c="d"/>
<b c="d"/>
</a>
</a>
</a>
the whole document is returned twice.
I assume either the query or the XMLoutput to be
invalid. The XQuery specs say (Section 3.2.1):
"""An attribute node may have an element node as its
parent, even though the attribute node is not a child
of the element node."""
Therefore my guess is that the query is valid and the
output is not.
## Comment 5191
Date: 2006-06-15 12:38:22 +0200
From: @peterboncz
Logged In: YES
user_id=591107
the problem has nothing to do with parent
the problem is that XPath specifies that the parent of an
attribute node is the element it is in. However, attribute
nodes are not children of the element node.
In DOM, attributed nodes are not children of elements.
pathfinder follows dom.
still, we allow these steps syntactically. as our item
oid-nunmbers for attributes are fantasy, the results of
feeding these as context nodes into staircase join are
fictional as well.
Jan R. has this been thought through?
## Comment 5192
Date: 2006-06-15 19:10:58 +0200
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
Like Peter said the problem here is that we do not support
steps originating from attributes. Thus the result is
clearly bogus (it interprets the IDs of the attributes as
pre values).
## Comment 5193
Date: 2006-06-15 22:09:00 +0200
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
I added a check in the head that prohibits attributes as
context nodes of path steps. This fixes the undefined behaviour.
Your queries now trigger unsupported feature errors. If this
is an important feature to someone - he/she might write an
feature request.
## Comment 5194
Date: 2006-06-19 09:34:07 +0200
From: @drstmane
Logged In: YES
user_id=572415
Re-opened as the now available tests fail with Peter's
revised checks/implementation:
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XQuery/attr_err2.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XQuery/attr_err2.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XQuery/attr_err3.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XQuery/attr_err3.err.00.html
## Comment 5195
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1459974 at http://sourceforge.net/support/tracker.php?aid=1459974
| XQuery: parent-step starting from attributes incorrect | https://api.github.com/repos/MonetDB/MonetDB/issues/904/comments | 0 | 2020-11-30T09:05:18Z | 2024-06-28T13:33:32Z | https://github.com/MonetDB/MonetDB/issues/904 | 753,293,474 | 904 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-27 04:38:22 +0200
From: tom <<dosithuyen>>
To: GDK devs <<bugs-common>>
Version: -- development
Last updated: 2007-04-06 04:20:01 +0200
## Comment 5186
Date: 2006-03-27 16:38:22 +0200
From: tom <<dosithuyen>>
I have two questions:
1)When I tried this query:
for $a in document
("directory.xml")/directory//doc_info[@update_time]
order by $a/@update_time
return $a/@doc_id
I encountered the error:
%>MapiClient -lx q1.xq
type error: [err:XPTY0004] Orderspec requires at most
one atomic item (given: string*).
I have tried using order by data($a/@doc_id) but it
still produced the same error.
2)When I tried this query:
for $a in document
("directory.xml")/directory//doc_info[@update_time]
return string($a/@doc_id)
I encountered the error:
%>MapiClient -lx q2.xq
type error: no variant of function fn:string accepts
the given argument type(s): attribute doc_id {
atomic* }*
type error: maybe you meant:
type error: fn:string () as string
type error: fn:string (item?) as string
type error: illegal arguments for function fn:string
And if I change string($a/@doc_id) to data
($a/@doc_id), it worked well.
Regards,
Tom
PS: I think they are not bugs but I don't know where
to post, then I post it here. Please move it to
appropriate place. Thanks you.
## Comment 5187
Date: 2006-04-05 15:16:57 +0200
From: Jens Teubner <<teubner>>
Logged In: YES
user_id=731390
Tom,
the problems you describe also fall in the category "static
typing". Here, our compiler's type inference for XPath
attribute steps is too weak to infer `attribute {}' for the
steps in your query.
Well, it was. I've committed a fix to our CVS repository
this morning that should provide better type inference at
least for the cases in your query.
You can access the fix by checking out the Pathfinder
compiler from CVS sources, or grabbing a nightly CVS
snapshot tarball from monetdb-xquery.org. Be aware, though,
that significant changes have been checked in in the past
weeks that might render the CVS version of MonetDB/XQuery
quite unstable.
The preferred communication path for user questions, by the
way, is the MonetDB users mailing list: http://lists.
sourceforge.net/lists/listinfo/monetdb-users.
Thanks for your interest in MonetDB/XQuery. Keep us posted
about the problems you encounter.
Jens
## Comment 5188
Date: 2007-04-06 04:20:01 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
---- Original comment by: sf-robot@users.sourceforge.net ----
Logged In: YES
user_id=1312539
Originator: NO
This Tracker item was closed automatically by the system. It was
previously set to a Pending status, and the original submitter
did not respond within 365 days (the time period specified by
the administrator of this Tracker).
## Comment 5189
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1459331 at http://sourceforge.net/support/tracker.php?aid=1459331
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-common@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
| MonetDB/Xquery with "order by" and "fn:string" | https://api.github.com/repos/MonetDB/MonetDB/issues/903/comments | 0 | 2020-11-30T09:05:16Z | 2024-06-27T11:09:44Z | https://github.com/MonetDB/MonetDB/issues/903 | 753,293,441 | 903 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-26 08:31:24 +0200
From: tom <<dosithuyen>>
To: GDK devs <<bugs-common>>
Version: -- development
Last updated: 2006-03-26 09:14:51 +0200
## Comment 5180
Date: 2006-03-26 20:31:24 +0200
From: tom <<dosithuyen>>
When I tried the query q3.xq (from Xmach-1, as I
attach below) using MapiClient, I encountered this
errorL
%> MapiClient -lx q3.xq
type error: no variant of function
local:DeepestFirstSection accepts the given argument
type(s): node*
type error: maybe you meant:
type error: local:DeepestFirstSection (node) as node
type error: illegal arguments for function
local:DeepestFirstSection
---q3.xq---
declare function local:DeepestFirstSection($e as node
()) as node()
{
if (empty($e/section2)) then $e
else local:DeepestFirstSection($e/section2[1])
};
local:DeepestFirstSection(doc("/root/Tom/XML/Xmach-
1/document.xml")/document/document2
[@doc_id="d6"]/chapter2[1]/section2[1])
I have tried to use element() instead of node() but
the same error happened.
Regards,
Tom
## Comment 5181
Date: 2006-03-26 21:11:43 +0200
From: Jens Teubner <<teubner>>
Logged In: YES
user_id=731390
Thanks for your interest in MonetDB/XQuery, Tom.
In contrast to most other XQuery implementations currently
around, MonetDB/XQuery provides XQuery's "static typing
feature". That is, at query compilation time, we infer type
information for each XQuery subexpression, check for the
type correctness of the query, and reject queries that are
incorrectly typed.
Path expressions in general could yield a *sequence* of XML
tree nodes with arbitrary length (type node()*). Your
implementation of DeepestFirstSection, however, allows only
exactly one node (type node()) as its input, and the type
check fails. (You might argue that the positional predicate
will ensure there is at most one node in the sequence, i.e.,
type node()?. For some internal reasons, however, our
compiler cannot yet detect that, hence, it will derive
node()*. Both will lead to an error message in your case,
though.)
Solutions:
(1) change your function to allow node()* as its input type
(bad style), or (much better)
(2) wrap function arguments into the function fn:exactly-
one() (e.g., `DeepestFirstSection($e/section2[1])'). This
will "silence" the static typechecker and defer the type
check into runtime. Analogously, there are also functions
fn:zero-or-one() and fn:one-or-more().
Once again, despite a minor issue about the specific case of
positional predicates (we're working on that :-)), this is a
actually a feature. If you're further interested, http://
www.w3.org/TR/xquery/dt-static-typing-feature might be
interesting for you.
If run into further problems, don't hesitate to ask again.
Jens
## Comment 5182
Date: 2006-03-26 21:11:51 +0200
From: @Teggy
Logged In: YES
user_id=642632
Hi Tom,
you're seeing this error messages because the static
typing in Pathfinder (MonetDB/XQuery's XQuery compiler) is
not yet as precise as it {c,s}hould be: your function
expects a single node, while the argument of the recursive
call is the outcome of a path expression which has type
node()*, i.e., a sequence of nodes.
Of course, the type inference phase of Pathfinder could
assign a more precise type to your path expression
$e/section2[1]: this precise type would be
(element section2)?
i.e., no or exactly one element of tag name section2. As I
said above, Pathfinder yields the type node()* instead we is
not as precise. (This will be improved in the near future.)
You can fix your function, though: rewrite it as
declare function local:DeepestFirstSection($e as node()) as
node()
{
if (empty($e/section2))
then $e
else
local:DeepestFirstSection(fn:exactly-one($e/section2[1]))
};
fn:exactly-one() checks -- at runtime -- that its argument
sequence contains exactly one item or no item at all (due to
your if-then-else, it is guaranteed that the runtime check
will succeed). At compile time, however, the compiler now
infers the type node() for the above application of
fn:exactly-one(). This is an acceptable input type to your
function.
So, you encountered an imprecision in Pathfinder's type
inference. Not a bug, though.
Regards,
--Torsten
## Comment 5183
Date: 2006-03-26 21:19:52 +0200
From: @Teggy
Logged In: YES
user_id=642632
Woops, I wrote ''... that its argument
sequence contains exactly one item or no item at all...''.
What I meant is ''... that its argument sequence contains
exactly one item...''.
--Torsten
## Comment 5184
Date: 2006-03-27 16:13:59 +0200
From: tom <<dosithuyen>>
Logged In: YES
user_id=1427893
Thank you very much for helping me! I am really interested
in exploring MonetDB/XQuery.
I have tried this:
declare function local:DeepestFirstSection($e as node()) as
node()
{
if (empty($e/section2))
then $e
else
local:DeepestFirstSection(fn:exactly-one($e/section2
[1]))
};
but it still produced same error. Therefore, I changed my
method to
declare function local:DeepestFirstSection($e as node()*)
as node()*
{
if (empty($e/section2))
then $e
else
local:DeepestFirstSection($e/section2[1])
};
and it worked.
Regards,
Tom
## Comment 5185
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1458865 at http://sourceforge.net/support/tracker.php?aid=1458865
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-common@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
| Does MonetDB/Xquery support node() and element()? | https://api.github.com/repos/MonetDB/MonetDB/issues/902/comments | 0 | 2020-11-30T09:05:13Z | 2024-06-28T13:34:20Z | https://github.com/MonetDB/MonetDB/issues/902 | 753,293,398 | 902 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-24 01:13:08 +0100
From: Jens Teubner <<teubner>>
To: @njnes
Version: -- development
Last updated: 2006-04-07 04:31:22 +0200
## Comment 5177
Date: 2006-03-24 13:13:08 +0100
From: Jens Teubner <<teubner>>
The HowToStart files are missing in the Pathfinder
distribution tarball.
Actually, why is there a "HowToStart", and an *empty*
README file? Wouldn't it make sense to just rename
HowToStart-PF into README, which would then
automatically be distributed with the tarball.
## Comment 5178
Date: 2006-04-07 16:31:22 +0200
From: @njnes
Logged In: YES
user_id=43556
The HowToStarts in the head version are in the EXTRA_DIST
variable, ie are now distributed. Also the README now points
to the howtos. We are using HowTo* files in all projects, ie
changing to use README is not prefered. These HowTos are
also the basis for the Howtos on the MonetDB website.
## Comment 5179
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1457701 at http://sourceforge.net/support/tracker.php?aid=1457701
| HowToStart files not in distribution tarball | https://api.github.com/repos/MonetDB/MonetDB/issues/901/comments | 0 | 2020-11-30T09:05:10Z | 2024-06-27T11:09:42Z | https://github.com/MonetDB/MonetDB/issues/901 | 753,293,354 | 901 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-23 04:17:58 +0100
From: Mark Lakata <<lakata>>
To: @njnes
Version: -- development
Last updated: 2006-03-23 11:33:30 +0100
## Comment 5174
Date: 2006-03-23 04:17:58 +0100
From: Mark Lakata <<lakata>>
The default configuration using
mel> flex --version
flex version 2.5.4
mel> bison --version
bison (GNU Bison) 1.875b
fails in buildtools/mel directory. Here is the error.
mel> nmake /f Makefile.msc
Microsoft (R) Program Maintenance Utility Version 7.10.3077
Copyright (C) Microsoft Corporation. All rights reserved.
cl -I. -GF -wd4273 -wd4102 -W3 -MD -nologo -Zi
/Femel.exe any_arg.obj atom_arg.obj atom.obj
atomops.obj bat_arg.obj builtin.obj command.obj
debug.obj depend.obj dependency.obj epilogue.obj
FileInfo.obj glue.obj html.obj iterator.obj
link.obj ListIterator.obj mel.obj mel.tab.obj
mel.yy.obj mil.obj module.obj object.obj
operator.obj ops.obj prelude.obj proto.obj
symbol.obj symtable.obj type_arg.obj use.obj
var_arg.obj /link /subsystem:console /NODEFAULTLIB:LIBC
mel.yy.obj : error LNK2019: unresolved external symbol
"int __cdecl isatty(int)" (?isatty@@YAHH@Z) referenced
in function "void __cdecl yy_init_buffer(struct
yy_buffer_state *,struct _iobuf *)"
(?yy_init_buffer@@YAXPAUyy_buffer_state@@PAU_iobuf@@@Z)
mel.exe : fatal error LNK1120: 1 unresolved externals
NMAKE : fatal error U1077: 'cl' : return code '0x2'
Stop.
mel>
I edited mel.ll manually and added this line to the
first line:
mel> head -1 mel.ll
%option never-interactive
and it compiled without error.
I am compiling using MSVC 2003 .Net for C++ (version 7)
in the cygwin environment.
## Comment 5175
Date: 2006-03-23 11:33:30 +0100
From: @njnes
Logged In: YES
user_id=43556
Thanks for the patch. Its applied on main branch now.
Does this solve all compiling problems etc on windows using
native (visual studio) with cygwin (ie without the not
maintained unixutils?)
## Comment 5176
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1456653 at http://sourceforge.net/support/tracker.php?aid=1456653
| windows build fails | https://api.github.com/repos/MonetDB/MonetDB/issues/900/comments | 0 | 2020-11-30T09:05:06Z | 2024-06-27T11:09:41Z | https://github.com/MonetDB/MonetDB/issues/900 | 753,293,315 | 900 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-21 12:34:45 +0100
From: @grobian
To: @sjoerdmullender
Version: -- development
Last updated: 2006-04-14 02:21:12 +0200
## Comment 5171
Date: 2006-03-21 12:34:45 +0100
From: @grobian
The bootstrap scripts used to bootstrap all modules
output errors to stderr instead of stderr. This leads
to unexpected abortion of the bootstrap script if
stdout is redirected to e.g. /dev/null.
I think it's a general rule of thumb to output (fatal)
errors to stderr instead of stdout.
Redirection could be done as in the attached patch.
## Comment 5172
Date: 2006-04-14 14:21:12 +0200
From: @sjoerdmullender
Logged In: YES
user_id=43607
Fixed.
## Comment 5173
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1455312 at http://sourceforge.net/support/tracker.php?aid=1455312
| bootstap script should output errors to stderr iso stdout | https://api.github.com/repos/MonetDB/MonetDB/issues/899/comments | 0 | 2020-11-30T09:05:03Z | 2024-06-27T11:09:40Z | https://github.com/MonetDB/MonetDB/issues/899 | 753,293,273 | 899 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-20 11:39:35 +0100
From: Joris van Rantwijk <<jvrantwijk>>
To: @njnes
Version: -- development
Last updated: 2007-02-19 05:34:15 +0100
## Comment 5163
Date: 2006-03-20 11:39:35 +0100
From: Joris van Rantwijk <<jvrantwijk>>
MonetDB does not close the socket connection to the
client after processing an XQuery. Every new connection
to the XQuery port opens a new socket which does not
get closed. When sending many queries to MonetDB using
MapiClient -lxquery, the server eventually runs out of
filehandles.
I tested this on Linux with MonetDB 4.10.3 and the CVS
XQuery_10 branch of Pathfinder; but the relevant code
has not changed since Pathfinder 0.10.2
From looking at the source code in pathfinder.mx, I
believe that it is also leaking stream buffer structures.
The attached patch fixes the problem for me. I'm not
sure if this is the right way to fix it though, since I
am very far from a MonetDb expert.
## Comment 5164
Date: 2006-03-21 12:28:51 +0100
From: @grobian
Logged In: YES
user_id=963970
Thanks for your patch!
This looks like code Niels made some changes to. I'm not
sure if your patch retains the behaviour that is envisioned,
so I guess he should say something about it.
## Comment 5165
Date: 2006-03-21 13:04:04 +0100
From: @grobian
Logged In: YES
user_id=963970
I just ran 11000 `MapiClient -lxquery` connects and
disconnects on the current development version without any
problem.
Could you try this on the current (development) CVS branch?
If not, I think this bug has already been fixed because the
code that handles the connections has been largerly rewritten.
## Comment 5166
Date: 2006-03-21 13:50:59 +0100
From: Joris van Rantwijk <<jvrantwijk>>
Logged In: YES
user_id=1480851
You are right. The development branch does not seem to have
this problem; only the stable branch.
## Comment 5167
Date: 2006-03-21 13:56:25 +0100
From: @grobian
Logged In: YES
user_id=963970
ok, thanks, will be fixed in x.12 releases
## Comment 5168
Date: 2006-11-29 10:58:54 +0100
From: @drstmane
Logged In: YES
user_id=572415
Originator: NO
Re-opened, as no test has been provided, yet.
User (submitter) and/or developer (assignee),
please supply a respective test,
or explain why a test is not required/possible.
## Comment 5169
Date: 2007-02-19 17:34:15 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Originator: NO
Added tests in the form of the smack00 and smack01 tests.
## Comment 5170
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1454407 at http://sourceforge.net/support/tracker.php?aid=1454407
| Mserver does not close XQuery sockets | https://api.github.com/repos/MonetDB/MonetDB/issues/898/comments | 0 | 2020-11-30T09:05:00Z | 2024-06-28T13:34:19Z | https://github.com/MonetDB/MonetDB/issues/898 | 753,293,228 | 898 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-20 02:40:59 +0100
From: @yzchang
To: XQuery devs <<bugs-xquery>>
Version: -- development
Last updated: 2006-03-20 03:55:19 +0100
## Comment 5159
Date: 2006-03-20 02:40:59 +0100
From: @yzchang
The following query (taken from the book "XQuery from
the Experts"):
let $x as xs:double := xs:double(1+2) return $x + $x
Results in:
$mapiclnt -lx query.xq
error in function application: at (1,25-1,40):
reference to undefined function `xs:double'
halted in ../../../compiler/semantics/functions.c
(check_fun_usage),
line 288
Has pathfinder incorrectly recognized the type casting
as a function reference, or a type casting like this is
not allowed (by XQuery, or by pathfinder)?
## Comment 5160
Date: 2006-03-20 11:37:24 +0100
From: Jens Teubner <<teubner>>
Logged In: YES
user_id=731390
There is no such thing as a `type cast' operator in XQuery.
What you were trying to use is a constructor function that
(unfortunately) is not implemented (yet) in Pathfinder.
## Comment 5161
Date: 2006-03-20 15:55:19 +0100
From: @yzchang
Logged In: YES
user_id=341633
Then this is more a feature than a bug, so I close it for now.
Thanks Jens for correcting me.
## Comment 5162
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1454150 at http://sourceforge.net/support/tracker.php?aid=1454150
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-xquery@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
| PF: constructor function recognized as function ref | https://api.github.com/repos/MonetDB/MonetDB/issues/897/comments | 0 | 2020-11-30T09:04:56Z | 2024-06-28T13:34:18Z | https://github.com/MonetDB/MonetDB/issues/897 | 753,293,201 | 897 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-17 12:43:07 +0100
From: thijs <<thijswes>>
To: Fabian Groffen <<fabian>>
Version: -- development
Last updated: 2007-01-12 05:03:30 +0100
## Comment 5154
Date: 2006-03-17 12:43:07 +0100
From: thijs <<thijswes>>
When XQuery output is printed in xml-noroot mode, the
command line is never returned.
Try for example:
echo '1'|MapiClient -lx -s xml-noroot
## Comment 5155
Date: 2006-03-21 14:21:08 +0100
From: @grobian
Logged In: YES
user_id=963970
The server did not send a terminating newline in xml-noroot
mode. I checked in a patch which seems to fix this. Please
reopen this bug if it doesn't work out for you.
## Comment 5156
Date: 2006-11-29 10:58:45 +0100
From: @drstmane
Logged In: YES
user_id=572415
Originator: NO
Re-opened, as no test has been provided, yet.
User (submitter) and/or developer (assignee),
please supply a respective test,
or explain why a test is not required/possible.
## Comment 5157
Date: 2007-01-12 17:03:30 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Originator: NO
Test added as tests/BugTracker noroot.SF-1452237
## Comment 5158
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1452237 at http://sourceforge.net/support/tracker.php?aid=1452237
| output mode xml-noroot never returns to command line | https://api.github.com/repos/MonetDB/MonetDB/issues/896/comments | 0 | 2020-11-30T09:04:53Z | 2024-06-27T11:09:37Z | https://github.com/MonetDB/MonetDB/issues/896 | 753,293,156 | 896 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-14 02:48:35 +0100
From: Jan Rittinger <<jan.rittinger>>
To: @njnes
Version: -- development
Last updated: 2006-05-12 07:25:38 +0200
## Comment 5148
Date: 2006-03-14 14:48:35 +0100
From: Jan Rittinger <<jan.rittinger>>
Seqbase of a bat changes when another bat is appended
(See line 80 of attached script).
## Comment 5149
Date: 2006-03-22 16:48:06 +0100
From: Pseudo user for Sourceforge import <<sfimport>>
---- Original comment by: kerschbaumers@users.sourceforge.net ----
Logged In: YES
user_id=1470010
The problem seems to happen if you use the pumped sum ({sum}) function (see
my attachment). If you comment out the call to sum, MonetDB prints the right
result.
## Comment 5150
Date: 2006-03-22 16:54:42 +0100
From: Pseudo user for Sourceforge import <<sfimport>>
---- Original comment by: kerschbaumers@users.sourceforge.net ----
Logged In: YES
user_id=1470010
Here is the MIL code:
var a := new (void, oid).seqbase(0@0);
var b := new (void, lng).seqbase(0@0);
a.append(1@0).append(2@0).append(1@0).append(2@0);
b.append(lng(1)).append(lng(2)).append(lng(1)).append(lng(1));
if you comment out the next line, the problem disappears
{sum}(b.reverse ().join (a).reverse ());
var y := new (lng, void).key (true).reverse ().seqbase (0@0);
var x := new (oid, lng);
x.append(lng(1));
x.append(lng(2));
y.append(x);
y.print();
quit();
## Comment 5151
Date: 2006-04-07 21:33:43 +0200
From: @njnes
Logged In: YES
user_id=43556
The general monet aggregate implementation (ie {}), does some
hackery on a temporary bat (called nested inside the
monet_setaggr function). This bat is properly destroyed using
BBPreclaim, but since the bat recycling this is now reused.
So the recycling now tests if bats with void types are
correctly varsized too.
## Comment 5152
Date: 2006-06-20 14:56:55 +0200
From: @njnes
Logged In: YES
user_id=43556
test exists
./tests/BugTracker/Tests/append_spoils_seqbase.SF-1449572.stable.out
## Comment 5153
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1449572 at http://sourceforge.net/support/tracker.php?aid=1449572
| append spoils seqbase | https://api.github.com/repos/MonetDB/MonetDB/issues/895/comments | 0 | 2020-11-30T09:04:50Z | 2024-06-27T11:09:37Z | https://github.com/MonetDB/MonetDB/issues/895 | 753,293,119 | 895 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-14 11:51:35 +0100
From: @sjoerdmullender
To: @sjoerdmullender
Version: -- development
Last updated: 2007-01-20 11:51:23 +0100
## Comment 5144
Date: 2006-03-14 11:51:35 +0100
From: @sjoerdmullender
It can happen that when you "make install" MonetDB and
then compile one of the other projects (sql or
pathfinder), make wants to rebuild some .h files in the
installation directory. This can happen when the .h
files that were generated by mel have the same time
stamp as the (also installed) corresponding .m file,
and make happens to read the directory in the wrong order.
If the install was done using a privileged user, the
attempt to rebuild the .h fails and hence the make fails.
## Comment 5145
Date: 2006-06-06 21:59:49 +0200
From: @grobian
Logged In: YES
user_id=963970
any (possible) solution here?
## Comment 5146
Date: 2007-01-20 23:51:23 +0100
From: @mlkersten
Logged In: YES
user_id=490798
Originator: NO
The BIGBANG has lead to a significant change
in configuration and build sequences.
It is unclear if the dependency errors
still hold.
Bug report closed until new signals of
erroneous behavior are reported.
## Comment 5147
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1449491 at http://sourceforge.net/support/tracker.php?aid=1449491
| BUILD: make dependency problem | https://api.github.com/repos/MonetDB/MonetDB/issues/894/comments | 0 | 2020-11-30T09:04:47Z | 2024-06-27T11:09:36Z | https://github.com/MonetDB/MonetDB/issues/894 | 753,293,083 | 894 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-10 02:02:18 +0100
From: @grobian
To: XQuery devs <<bugs-xquery>>
Version: -- development
Last updated: 2006-12-17 05:15:20 +0100
## Comment 5139
Date: 2006-03-10 14:02:18 +0100
From: @grobian
From the Italian guy:
in terminal 1:
---
(pegasus:~/vtmp) fabian% cat short.mil
module(mapi);
module(sql_server);
module(pathfinder);
var allow_foreign_connections := false;
mapi_register(mil_frontend());
mapi_register(sql_frontend());
mapi_register(xquery_frontend(lng(0)));
fork(mapi_listen(
int(monet_environment.find("mapi_port")),
str(nil),
5,
allow_foreign_connections)
);
(pegasus:~/vtmp) fabian% $INSTALL_DIR/bin/Mserver short.mil
---
in terminal 2: (factor1.xml is the xmark file from the
assets website)
---
(pegasus:~) fabian% echo
'shred_doc("file:///home/fabian/vtmp/factor1.xml",
"factor1.xml");' | MapiClient -lmil
(pegasus:~) fabian% echo 'let $auction :=
doc("factor1.xml") return for $b in
$auction/site/open_auctions/open_auction return
<increase>{$b/bidder[1]/increase/text()}</increase>' |
MapiClient -lxq
... (some output)
---
in terminal 1:
---
^D (terminate Mserver)
(pegasus:~/vtmp) fabian% $INSTALL_DIR/bin/Mserver short.mil
Monet Database Server V4.11.3
Copyright (c) 1993-2006, CWI. All rights reserved.
Compiled for x86_64-redhat-linux-gnu/64bit with 64bit
OIDs; dynamically linked.
Visit http://monetdb.cwi.nl/ for further information.
!WARNING: BBPincref: range error 52
!WARNING: BBPincref: range error 53
!WARNING: BBPincref: range error 54
!WARNING: BBPincref: range error 173
!WARNING: BBPincref: range error 176
!WARNING: BBPincref: range error 178
!WARNING: BBPincref: range error 58
!WARNING: BBPincref: range error 59
!WARNING: BBPincref: range error 60
!WARNING: BBPincref: range error 61
!WARNING: BBPincref: range error 180
!WARNING: BBPincref: range error 183
!WARNING: BBPincref: range error 185
!WARNING: BBPincref: range error 65
!WARNING: BBPincref: range error 66
!WARNING: BBPincref: range error 67
!WARNING: BBPincref: range error 68
!WARNING: BBPincref: range error 69
!WARNING: BBPincref: range error 70
!WARNING: BBPincref: range error 71
!WARNING: BBPincref: range error 72
!WARNING: BBPincref: range error 187
!WARNING: BBPincref: range error 190
!WARNING: BBPincref: range error 192
!WARNING: BBPincref: range error 194
!WARNING: BBPincref: range error 196
!WARNING: BBPincref: range error 198
!WARNING: BBPincref: range error 200
!WARNING: BBPincref: range error 80
!WARNING: BBPincref: range error 81
!WARNING: BBPincref: range error 82
!WARNING: BBPincref: range error 83
!WARNING: BBPincref: range error 84
!WARNING: BBPincref: range error 85
!WARNING: BBPincref: range error 86
!WARNING: BBPincref: range error 87
!WARNING: BBPincref: range error 88
!WARNING: BBPincref: range error 89
!WARNING: BBPincref: range error 202
!WARNING: BBPincref: range error 205
!WARNING: BBPincref: range error 207
!WARNING: BBPincref: range error 209
!WARNING: BBPincref: range error 211
!WARNING: BBPincref: range error 213
!WARNING: BBPincref: range error 215
!WARNING: BBPincref: range error 217
!WARNING: BBPincref: range error 219
!WARNING: BBPincref: range error 99
!WARNING: BBPincref: range error 100
!WARNING: BBPincref: range error 101
!WARNING: BBPincref: range error 102
!WARNING: BBPincref: range error 103
!WARNING: BBPincref: range error 104
!WARNING: BBPincref: range error 221
!WARNING: BBPincref: range error 224
!WARNING: BBPincref: range error 226
!WARNING: BBPincref: range error 228
!WARNING: BBPincref: range error 230
!WARNING: BBPincref: range error 110
!WARNING: BBPincref: range error 111
!WARNING: BBPincref: range error 112
!WARNING: BBPincref: range error 113
!WARNING: BBPincref: range error 114
!WARNING: BBPincref: range error 241
!WARNING: BBPincref: range error 244
!WARNING: BBPincref: range error 246
!WARNING: BBPincref: range error 248
!WARNING: BBPincref: range error 119
!WARNING: BBPincref: range error 120
!WARNING: BBPincref: range error 121
!WARNING: BBPincref: range error 122
!WARNING: BBPincref: range error 123
!WARNING: BBPincref: range error 232
!WARNING: BBPincref: range error 235
!WARNING: BBPincref: range error 237
!WARNING: BBPincref: range error 239
!WARNING: BBPincref: range error 128
!WARNING: BBPincref: range error 129
!WARNING: BBPincref: range error 130
!WARNING: BBPincref: range error 131
!WARNING: BBPincref: range error 132
!WARNING: BBPincref: range error 133
!WARNING: BBPincref: range error 134
!WARNING: BBPincref: range error 135
!WARNING: BBPincref: range error 136
!WARNING: BBPincref: range error 250
!WARNING: BBPincref: range error 255
!WARNING: BBPincref: range error 257
!WARNING: BBPincref: range error 259
!WARNING: BBPincref: range error 261
!WARNING: BBPincref: range error 263
!WARNING: BBPincref: range error 265
!WARNING: BBPincref: range error 253
!WARNING: BBPincref: range error 145
!WARNING: BBPincref: range error 146
!WARNING: BBPincref: range error 147
!WARNING: BBPincref: range error 148
!WARNING: BBPincref: range error 149
!WARNING: BBPincref: range error 150
!WARNING: BBPincref: range error 151
!WARNING: BBPincref: range error 267
!WARNING: BBPincref: range error 270
!WARNING: BBPincref: range error 272
!WARNING: BBPincref: range error 274
!WARNING: BBPincref: range error 276
!WARNING: BBPincref: range error 278
!WARNING: BBPincref: range error 158
!WARNING: BBPincref: range error 159
!WARNING: BBPincref: range error 160
!WARNING: BBPincref: range error 161
!WARNING: BBPincref: range error 162
!WARNING: BBPincref: range error 163
!WARNING: BBPincref: range error 164
!WARNING: BBPincref: range error 165
!WARNING: BBPincref: range error 280
!WARNING: BBPincref: range error 283
!WARNING: BBPincref: range error 285
!WARNING: BBPincref: range error 287
!WARNING: BBPincref: range error 289
!WARNING: BBPincref: range error 291
!WARNING: BBPincref: range error 293
!WARNING: BBPincref: range error 295
!WARNING: BBPincref: range error 427
!WARNING: BBPincref: range error 525
!WARNING: BBPincref: range error 428
!WARNING: BBPincref: range error 527
!WARNING: BBPincref: range error 429
!WARNING: BBPincref: range error 529
!WARNING: BBPincref: range error 430
!WARNING: BBPincref: range error 531
!WARNING: BBPincref: range error 431
!WARNING: BBPincref: range error 533
!WARNING: BBPincref: range error 432
!WARNING: BBPincref: range error 535
!WARNING: BBPincref: range error 433
!WARNING: BBPincref: range error 537
!WARNING: BBPincref: range error 426
!WARNING: BBPincref: range error 440
!WARNING: BBPincref: range error 552
!WARNING: BBPincref: range error 441
!WARNING: BBPincref: range error 554
!WARNING: BBPincref: range error 439
!WARNING: BBPincref: range error 443
!WARNING: BBPincref: range error 559
!WARNING: BBPincref: range error 444
!WARNING: BBPincref: range error 561
!WARNING: BBPincref: range error 445
!WARNING: BBPincref: range error 563
!WARNING: BBPincref: range error 446
!WARNING: BBPincref: range error 565
!WARNING: BBPincref: range error 447
!WARNING: BBPincref: range error 567
!WARNING: BBPincref: range error 448
!WARNING: BBPincref: range error 569
!WARNING: BBPincref: range error 449
!WARNING: BBPincref: range error 571
!WARNING: BBPincref: range error 450
!WARNING: BBPincref: range error 573
!WARNING: BBPincref: range error 451
!WARNING: BBPincref: range error 575
!WARNING: BBPincref: range error 452
!WARNING: BBPincref: range error 577
!WARNING: BBPincref: range error 442
!WARNING: BBPincref: range error 454
!WARNING: BBPincref: range error 590
!WARNING: BBPincref: range error 455
!WARNING: BBPincref: range error 592
!WARNING: BBPincref: range error 456
!WARNING: BBPincref: range error 594
!WARNING: BBPincref: range error 457
!WARNING: BBPincref: range error 596
!WARNING: BBPincref: range error 453
!WARNING: BBPincref: range error 459
!WARNING: BBPincref: range error 603
!WARNING: BBPincref: range error 460
!WARNING: BBPincref: range error 605
!WARNING: BBPincref: range error 458
!WARNING: BBPincref: range error 462
!WARNING: BBPincref: range error 610
!WARNING: BBPincref: range error 463
!WARNING: BBPincref: range error 612
!WARNING: BBPincref: range error 464
!WARNING: BBPincref: range error 614
!WARNING: BBPincref: range error 461
!WARNING: BBPincref: range error 466
!WARNING: BBPincref: range error 620
!WARNING: BBPincref: range error 467
!WARNING: BBPincref: range error 622
!WARNING: BBPincref: range error 468
!WARNING: BBPincref: range error 624
!WARNING: BBPincref: range error 469
!WARNING: BBPincref: range error 626
!WARNING: BBPincref: range error 470
!WARNING: BBPincref: range error 628
!WARNING: BBPincref: range error 465
!WARNING: BATdescriptor: range error 69
!WARNING: BATdescriptor: range error 69
!ERROR: BATSIGcrash: Mserver internal error
(Segmentation fault), please restart.
!ERROR: (One potential cause could be that your disk
might be full...)
(pegasus:~/vtmp) fabian%
---
## Comment 5140
Date: 2006-03-10 14:41:49 +0100
From: @grobian
Logged In: YES
user_id=963970
Note: without sql (i.e. removing module(sql) and
mapi_register(sql_frontend()) from the startup script) the
crash does not appear.
## Comment 5141
Date: 2006-08-17 00:12:53 +0200
From: @mlkersten
Logged In: YES
user_id=490798
Is this bug still valid?
## Comment 5142
Date: 2006-12-17 17:15:20 +0100
From: @peterboncz
Logged In: YES
user_id=591107
Originator: NO
this is related to the interaction between SQL and XQuery in <= .12
this was never officially supported. Only from .14 on this will be possible.
## Comment 5143
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1447195 at http://sourceforge.net/support/tracker.php?aid=1447195
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-xquery@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
| PF: crash server with XMark | https://api.github.com/repos/MonetDB/MonetDB/issues/893/comments | 0 | 2020-11-30T09:04:44Z | 2024-06-28T13:34:17Z | https://github.com/MonetDB/MonetDB/issues/893 | 753,293,042 | 893 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-10 12:49:09 +0100
From: @grobian
To: GDK devs <<bugs-common>>
Version: -- development
Last updated: 2006-06-07 01:36:18 +0200
## Comment 5134
Date: 2006-03-10 12:49:09 +0100
From: @grobian
Consider the following script:
---
var allow_foreign_connections := false;
module(sql_server);
sql_server_start(allow_foreign_connections);
print("SQL started");
module(pathfinder);
xquery_server_start(allow_foreign_connections);
print("XQuery started");
module(mapi);
mapi_start(allow_foreign_connections);
print("MAPI started");
---
Starting Mserver and pasting line by line the contents
of this script in the console, results in a flawless
cooperation of Mserver. When the script is fed to the
Mserver as argument (not with a pipe (<)!) OS binding
errors occur for port 50000. Example (the output is
random):
---
[ "SQL started" ]
[ "XQuery started" ]
[ "MAPI started" ]
MonetDB>!ERROR: MAPIlisten:binding to stream socket
(50000) failed
!OS: Address already in use
---
Using Mserver < myscript.mil the problem does not
happen. Sometimes two bind errors appear. It looks as
if the 50000 port is always bound to first, then
released if another port is necessary. When running
from a script, releasing of the port isn't fast enough.
Maybe we can use the "bind, also when in use" flag to
avoid the problem here?
## Comment 5135
Date: 2006-06-06 22:09:18 +0200
From: @grobian
Logged In: YES
user_id=963970
simply running the script is enough to just crash monetdb
with ease.
Is there no way to protect from this? This is quite
critical by the way as I need this functionality in Gentoo.
## Comment 5136
Date: 2006-06-07 09:09:47 +0200
From: @grobian
Logged In: YES
user_id=963970
Simply change your call to something like
module(sql_server);
module(pathfinder);
mapi_start(allow_foreign_connections);
mapi_register(sql_frontend());
mapi_register(xquery_frontend());
Niels
(pegasus:current/monetdb/tests) fabian% cat loadcrash.mil
var allow_foreign_connections := false;
module(sql_server);
module(pathfinder);
mapi_start(allow_foreign_connections);
mapi_register(sql_frontend());
mapi_register(xquery_frontend());
(pegasus:current/monetdb/tests) fabian% run_monet --set
mapi_port=50002 < loadcrash.mil
Monet Database Server V4.11.3
Copyright (c) 1993-2006, CWI. All rights reserved.
Compiled for x86_64-redhat-linux-gnu/64bit with 64bit
OIDs; dynamically linked.
Visit http://monetdb.cwi.nl/ for further information.
MonetDB>MonetDB>MonetDB>MonetDB>!WARNING: BBPdecref: range
error 273
!WARNING: interpret_params: range error 273
!ERROR: Myyclean: BAT 273 OVERWRITTEN (stamp -168 != 0)
(crash)
(after a clean of my dbfarm the warning and error messages
disappear, but the crash remains. Cannot debug it as I
don't know how to pipe a script into a process debugged by gdb)
(note that it doesn't crash if the script is given as
argument, only on pipe)
## Comment 5137
Date: 2006-06-07 13:36:18 +0200
From: @grobian
Logged In: YES
user_id=963970
Probably user error in the last case. The new script works
as script, which is the only way to use it in a
monet_daemon=yes invironment. The new sequence as provided
by Niels is the way to do it.
## Comment 5138
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1447164 at http://sourceforge.net/support/tracker.php?aid=1447164
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-common@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
| Starting multiple server backends too fast result in bind er | https://api.github.com/repos/MonetDB/MonetDB/issues/892/comments | 0 | 2020-11-30T09:04:40Z | 2024-06-28T13:34:17Z | https://github.com/MonetDB/MonetDB/issues/892 | 753,292,994 | 892 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-09 02:36:52 +0100
From: thijs <<thijswes>>
To: Stefan Manegold <<Stefan.Manegold>>
Version: -- development
Last updated: 2006-06-14 12:12:47 +0200
## Comment 5128
Date: 2006-03-09 14:36:52 +0100
From: thijs <<thijswes>>
when a call to fn:doc is inside a nested series of
element (or attribute) constructors, it fails.
Suppose a file "test.xml" exists and is shredded (can
be any valid xml file). Then, a query with a fn:doc
inside nested element constructors produce a mposjoin
error:
element c {element d {fn:doc("test.xml")}}
gives
ERROR: interpret: no matching MIL operator to
'mposjoin(oid, BAT[oid,oid], BAT[oid,BAT])'.
MAYBE YOU MEAN:
mposjoin(BAT[oid,oid], oid, BAT[oid,BAT]) :
BAT[void,any]
mposjoin(BAT[oid,oid], BAT[oid,oid],
BAT[oid,BAT]) : BAT[void,any]
A query with a single element constructor gives the
expected result:
element c {fn:doc("test.xml")}
gives
<c><a>
<b/>
</a></c>
## Comment 5129
Date: 2006-06-04 17:28:41 +0200
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
The problem here is the following line:
var attr_item := attr.leftfetchjoin(item);
where attr is an empty (void, oid) bat and item is a
constant 0@0. It results in the constant 0@0 instead of an
empty bat...
So it seems to be once again something related to fake bats
and constants...
## Comment 5130
Date: 2006-06-06 21:55:11 +0200
From: @grobian
Logged In: YES
user_id=963970
Is there a (meta) bug we can let this one depend on?
## Comment 5131
Date: 2006-06-14 00:12:46 +0200
From: @peterboncz
Logged In: YES
user_id=591107
this should be fixed now
## Comment 5132
Date: 2006-06-20 11:13:56 +0200
From: @peterboncz
Logged In: YES
user_id=591107
added test script in tests/BugTracker/
## Comment 5133
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1446426 at http://sourceforge.net/support/tracker.php?aid=1446426
| fn:doc() call in nested element construction fails | https://api.github.com/repos/MonetDB/MonetDB/issues/891/comments | 0 | 2020-11-30T09:04:37Z | 2024-06-28T13:34:16Z | https://github.com/MonetDB/MonetDB/issues/891 | 753,292,954 | 891 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-06 01:44:02 +0100
From: @grobian
To: @njnes
Version: -- development
Last updated: 2006-04-07 05:00:53 +0200
## Comment 5124
Date: 2006-03-06 13:44:02 +0100
From: @grobian
On systems with netcdf available (amor and volans for
instance) make of MonetDB 4.11.3 (daily build tarbal)
fails:
test -e .libs || mkdir -p .libs
test -e .libs/mnetcdf.mil || ln -s ../mnetcdf.mil
.libs/mnetcdf.mil
make all-am
make[7]: Entering directory
`/var/tmp/MonetDB-4.11.3/build/src/modules/mnetcdf'
test -e mnetcdf.mil || ln -s
../../../../src/modules/mnetcdf/mnetcdf.mil mnetcdf.mil
ln: `mnetcdf.mil': File exists
make[7]: *** [mnetcdf.mil] Error 1
make[7]: Leaving directory
`/var/tmp/MonetDB-4.11.3/build/src/modules/mnetcdf'
make[6]: *** [all] Error 2
make[6]: Leaving directory
`/var/tmp/MonetDB-4.11.3/build/src/modules/mnetcdf'
make[5]: *** [all-recursive] Error 1
make[5]: Leaving directory
`/var/tmp/MonetDB-4.11.3/build/src/modules'
make[4]: *** [all] Error 2
make[4]: Leaving directory
`/var/tmp/MonetDB-4.11.3/build/src/modules'
make[3]: *** [all-recursive] Error 1
make[3]: Leaving directory
`/var/tmp/MonetDB-4.11.3/build/src'
make[2]: *** [all] Error 2
make[2]: Leaving directory
`/var/tmp/MonetDB-4.11.3/build/src'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/var/tmp/MonetDB-4.11.3/build'
make: *** [all] Error 2
configuring with --without-netcdf "solves" the problem.
Or modern boxen (such as pegasus) don't have netcdf,
so the problem doesn't arise.
## Comment 5125
Date: 2006-03-09 14:53:42 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
It seems to me the system where the tarball is made should
have netcdf installed.
Alternatively, we should remove mnetcdf completely, but I
guess not everyone will like that option...
## Comment 5126
Date: 2006-04-07 17:00:53 +0200
From: @njnes
Logged In: YES
user_id=43556
make dist now fails if netcdf is missing, so any tarball
should have the proper netcdf.mil files.
## Comment 5127
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1444091 at http://sourceforge.net/support/tracker.php?aid=1444091
| Compilation fails if netcdf is available | https://api.github.com/repos/MonetDB/MonetDB/issues/890/comments | 0 | 2020-11-30T09:04:33Z | 2024-06-27T11:09:32Z | https://github.com/MonetDB/MonetDB/issues/890 | 753,292,913 | 890 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-03-04 12:51:11 +0100
From: @grobian
To: Fabian Groffen <<fabian>>
Version: -- development
Last updated: 2006-03-04 03:24:17 +0100
## Comment 5121
Date: 2006-03-04 12:51:11 +0100
From: @grobian
(pegasus:~) fabian% gdb --args
$MONETDB_PREFIX/bin/MapiClient -lsql -uvoc
GNU gdb Red Hat Linux (6.3.0.0-1.84rh)
Copyright 2004 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public
License, and you are
welcome to change it and/or distribute copies of it
under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB. Type "show
warranty" for details.
This GDB was configured as
"x86_64-redhat-linux-gnu"...Using host libthread_db
library "/lib64/libthread_db.so.1".
(gdb) r
Starting program:
/net/pegasus.ins.cwi.nl/export/scratch1/fabian/monetdb/current/program-x86_64/bin/MapiClient
-lsql -uvoc
Password (monetdb):
Program received signal SIGSEGV, Segmentation fault.
0x00002aaaaaab2a98 in mapi_start_talking (mid=0x50b8d0)
at
/ufs/fabian/scratch/monetdb/current/monetdb/src/mapi/clients/C/Mapi.mx:2121
2121 hdl->result->errorstr =
NULL; /* clear these so errorstr doesn't get freed */
(gdb) bt
0 0x00002aaaaaab2a98 in mapi_start_talking (mid=0x50b8d0)
at
/ufs/fabian/scratch/monetdb/current/monetdb/src/mapi/clients/C/Mapi.mx:2121
1 0x00002aaaaaab2dbf in mapi_reconnect (mid=0x50b8d0)
at
/ufs/fabian/scratch/monetdb/current/monetdb/src/mapi/clients/C/Mapi.mx:2167
2 0x00002aaaaaab2e1c in mapi_connect (host=0x50b720
"localhost", port=50000,
username=0x7fffffffa1af "voc", password=0x50c260
"monetdb",
lang=0x408ee3 "sql")
at
/ufs/fabian/scratch/monetdb/current/monetdb/src/mapi/clients/C/Mapi.mx:2180
3 0x0000000000406a7a in main (argc=3,
argv=0x7fffffff8808)
at
/ufs/fabian/scratch/monetdb/current/monetdb/src/mapi/clients/C/MapiClient.mx:1413
(gdb) q
The program is running. Exit anyway? (y or n) y
Sidenote is that the server also crashes at this point,
so I think it only happens if the server doesn't send
anything back after the challenge.
## Comment 5122
Date: 2006-03-04 15:24:17 +0100
From: @grobian
Logged In: YES
user_id=963970
hdl->result was not checked often enough for not NULLness.
Not testable, as I fixed it in the server also.
## Comment 5123
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1443014 at http://sourceforge.net/support/tracker.php?aid=1443014
| MapiClient crashes on invalid credentials (M5) | https://api.github.com/repos/MonetDB/MonetDB/issues/889/comments | 0 | 2020-11-30T09:04:30Z | 2024-06-27T11:09:31Z | https://github.com/MonetDB/MonetDB/issues/889 | 753,292,873 | 889 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-28 10:31:12 +0100
From: @grobian
To: @njnes
Version: -- development
Last updated: 2006-03-07 08:11:10 +0100
## Comment 5117
Date: 2006-02-28 10:31:12 +0100
From: @grobian
From the VOC demo:
voc-> create table test (
voc=( id int,
voc=( data varchar(30)
voc=( );
Error: syntax error, unexpected DATA, expecting CHECK
or FOREIGN or PRIMARY or UNIQUE in: "create table test
( id int, data"
We should either fix the demo, or add this keyword to
the list of allowed "keywords taken as identifiers"
## Comment 5118
Date: 2006-03-07 11:46:17 +0100
From: @njnes
Logged In: YES
user_id=43556
Fixed (added DATA to the allowed identifiers reserved-word
list). Todo add test script (VOC?)
## Comment 5119
Date: 2006-03-07 20:11:10 +0100
From: @grobian
Logged In: YES
user_id=963970
meltdown: TEST ADDED
sql/src/test/BugTracker/Tests/create_column_name_data.SF-1440214.*
## Comment 5120
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1440214 at http://sourceforge.net/support/tracker.php?aid=1440214
| SQL: DATA is reserved keyword | https://api.github.com/repos/MonetDB/MonetDB/issues/888/comments | 0 | 2020-11-30T09:04:27Z | 2024-06-27T11:09:30Z | https://github.com/MonetDB/MonetDB/issues/888 | 753,292,842 | 888 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-23 06:13:11 +0100
From: @MarcinZukowski
To: GDK devs <<bugs-common>>
Version: -- development
Last updated: 2006-06-07 01:33:47 +0200
## Comment 5111
Date: 2006-02-23 18:13:11 +0100
From: @MarcinZukowski
MonetDB doesn't free all of the data it allocates.
It makes debugging MonetDB-based applications (like
X100) hardish, as e.g. valgrind output is cluttered by
the MonetDB-related warnings.
Example for a simple query where we just load a "unix"
module:
272 bytes in 1 blocks are possibly lost in loss rec
ord 10 of 16
at 0x4905DCD: calloc (vg_replace_malloc.c:279)
by 0x348E00F8E3: _dl_allocate_tls (in /lib64/ld-
2.3.5.so)
by 0x348EF063B4: pthread_create@@GLIBC_2.2.5 (in
/lib64/libpthread-2.3.5.so)
by 0x5A129C5: MT_create_thread (gdk_system.mx:39
2)
by 0x5915BFB: GDKinit (gdk_utils.mx:1475)
by 0x401DF1: Monet_init (Mserver.mx:172)
by 0x40262D: main (Mserver.mx:397)
792 (528 direct, 264 indirect) bytes in 2 blocks ar
e definitely lost in loss record 12 of 16
at 0x4904A66: malloc (vg_replace_malloc.c:149)
by 0x591480A: GDKmallocmax (gdk_utils.mx:1070)
by 0x5914995: GDKmalloc (gdk_utils.mx:1091)
by 0x4A3A3CB: yychunkalloc (yytree.mx:159)
by 0x4A3A5FD: yycopy (yytree.mx:187)
by 0x4A3A86C: yycopy (yytree.mx:215)
by 0x4A3B4C4: Myylogout (yytree.mx:479)
by 0x4A8519E: yyparse (monet_parse.y:42)
by 0x4A3E049: parseClient (monet_client.mx:715)
by 0x4A18143: interpret_str (monet_interpreter.m
x:244)
by 0x4A159B3: monet_eval (monet.mx:599)
by 0x4A15BB5: monet_exec (monet.mx:634)
by 0x4027C7: static_module_init (initmodules.mx:
35)
by 0x402666: main (Mserver.mx:404)
241,944 bytes in 12 blocks are possibly lost in los
s record 15 of 16
at 0x4905EC2: realloc (vg_replace_malloc.c:306)
by 0x5914CF9: GDKreallocmax (gdk_utils.mx:1191)
by 0x588F7AA: HEAPextend (gdk_heap.mx:202)
by 0x591CB8E: strPut (gdk_atoms.mx:1799)
by 0x5973069: BUNins (gdk_bat.mx:862)
by 0x4A81E68: TBL_Install_Function (monet_tbl.mx
:2999)
by 0x4A81AE6: TBL_Install_Functions (monet_tbl.m
x:2940)
by 0x4F6FA9E: _builtin_Module_Install (builtin.g
lue.c:760)
by 0x4A77808: TBL_M_load (monet_tbl.mx:1546)
by 0x4A78ADB: TBL_loadmod (monet_tbl.mx:1812)
by 0x4A3E28C: moduleClient (monet_client.mx:781)
by 0x4A85442: yyparse (monet_parse.y:109)
by 0x4A3E049: parseClient (monet_client.mx:715)
by 0x4A18143: interpret_str (monet_interpreter.m
x:244)
by 0x4A159B3: monet_eval (monet.mx:599)
by 0x4A15BB5: monet_exec (monet.mx:634)
by 0x4026E7: static_module_init (initmodules.mx:
35)
by 0x402666: main (Mserver.mx:404)
714,292 bytes in 2,894 blocks are possibly lost in
loss record 16 of 16
at 0x4904A66: malloc (vg_replace_malloc.c:149)
by 0x591480A: GDKmallocmax (gdk_utils.mx:1070)
by 0x5914995: GDKmalloc (gdk_utils.mx:1091)
by 0x5882EA3: BBPinithash (gdk_bbp.mx:362)
by 0x588424A: BBPinit (gdk_bbp.mx:667)
by 0x5915B6C: GDKinit (gdk_utils.mx:1465)
by 0x401DF1: Monet_init (Mserver.mx:172)
by 0x40262D: main (Mserver.mx:397)
## Comment 5112
Date: 2006-02-23 19:29:09 +0100
From: @drstmane
Logged In: YES
user_id=572415
I guess this refers to "MonetDB CVS Head", not "buildtools
CVS Head" ...
## Comment 5113
Date: 2006-03-01 15:42:40 +0100
From: @MarcinZukowski
Logged In: YES
user_id=607094
If someone else has a similar problem - valgrind allows
specifying error suppression with a --suppressions=FILE
flag. To get the file one can use --gen-suppressions=all and
then edit the output to contain suppressions.
I attach an example MONET.SUP file I now use.
## Comment 5114
Date: 2006-06-07 13:33:47 +0200
From: @grobian
Logged In: YES
user_id=963970
Niels claims he fixed this. Please reopen if the problem
persists.
## Comment 5115
Date: 2006-06-08 13:56:52 +0200
From: @MarcinZukowski
Logged In: YES
user_id=607094
I still have plenty of unfreed memory with just running
Mserver in valgrind:
valgrind -v -v -v --trace-children=yes --leak-check=yes
--tool=memcheck --log-file=MS --num-callers=48
--error-limit=no --alignment=8 --show-reachable=yes Mserver
So I don't think it's fully fixed, maybe a bit better, but
not fixed.
## Comment 5116
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1437559 at http://sourceforge.net/support/tracker.php?aid=1437559
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-common@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
| Minor memory leaks in MonetDB | https://api.github.com/repos/MonetDB/MonetDB/issues/887/comments | 0 | 2020-11-30T09:04:24Z | 2024-06-27T11:09:29Z | https://github.com/MonetDB/MonetDB/issues/887 | 753,292,798 | 887 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-23 11:25:22 +0100
From: @sjoerdmullender
To: @sjoerdmullender
Version: -- development
Last updated: 2006-02-23 04:58:36 +0100
## Comment 5107
Date: 2006-02-23 11:25:22 +0100
From: @sjoerdmullender
Inspired by a failure in pathfinder:
mil>c.print();
-----------------
t h name
void oid type
-----------------
[ 0@0, 1@0 ]
[ 1@0, 1@0 ]
mil>d.print();
-----------------
t h name
void void type
-----------------
[ 0@0, 1@0 ]
[ 1@0, 2@0 ]
mil>[<](c,d).print();
MAPI = anonymous@localhost:50000
QUERY = [<](c,d).print();
ERROR = !ERROR: interpret: no matching MIL operator to
'[<](BAT[oid,oid], BAT[oid,oid])'.
!MAYBE YOU MEAN:
! <(timestamp, timestamp) : bit
! <(daytime, daytime) : bit
! <(date, date) : bit
! <(str, str) : bit
! <(lng, lng) : bit
! <(dbl, dbl) : bit
! <(flt, flt) : bit
! <(oid, oid) : bit
! <(int, int) : bit
! <(sht, sht) : bit
! <(chr, chr) : bit
! <(any::1, any::1) : bit
!ERROR: interpret_params: print(param 1):
evaluation error.
mil>
## Comment 5108
Date: 2006-02-23 11:32:32 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Test added:
tests/BugTracker/Tests/multiplex_void_oid.SF-1437319.*
## Comment 5109
Date: 2006-02-23 16:58:35 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Fixed in CVS.
## Comment 5110
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1437319 at http://sourceforge.net/support/tracker.php?aid=1437319
| multiplex function void/oid parameter matching not correct | https://api.github.com/repos/MonetDB/MonetDB/issues/886/comments | 0 | 2020-11-30T09:04:20Z | 2024-06-27T11:09:28Z | https://github.com/MonetDB/MonetDB/issues/886 | 753,292,761 | 886 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-23 10:45:29 +0100
From: Jan Rittinger <<jan.rittinger>>
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-02-24 04:00:09 +0100
## Comment 5103
Date: 2006-02-23 10:45:29 +0100
From: Jan Rittinger <<jan.rittinger>>
nlthetajoin a feature that's incomplete :(
Try:
MonetDB>var a := bat(int,int).insert(1,1).insert(1,1);
MonetDB>module("pathfinder");
MonetDB>nlthetajoin(a,a,EQ,4LL).print();
-----------------
h t name
int int type
-----------------
[ 1, 1 ]
[ 1, 1 ]
[ 1, 1 ]
[ 1, 1 ]
MonetDB>nlthetajoin(a.tmark(0@0), a.mark(0@0), EQ,
4LL).print();
-----------------
h t name
oid oid type
-----------------
## Comment 5104
Date: 2006-02-24 16:00:09 +0100
From: @peterboncz
Logged In: YES
user_id=591107
fixed
## Comment 5105
Date: 2006-06-20 11:12:46 +0200
From: @peterboncz
Logged In: YES
user_id=591107
added test script in tests/BugTracker
## Comment 5106
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1437299 at http://sourceforge.net/support/tracker.php?aid=1437299
| nlthetajoin incomplete | https://api.github.com/repos/MonetDB/MonetDB/issues/885/comments | 0 | 2020-11-30T09:04:17Z | 2024-06-27T11:09:27Z | https://github.com/MonetDB/MonetDB/issues/885 | 753,292,732 | 885 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-22 01:32:55 +0100
From: @sjoerdmullender
To: @njnes
Version: -- development
Last updated: 2006-03-08 02:47:58 +0100
## Comment 5099
Date: 2006-02-22 13:32:55 +0100
From: @sjoerdmullender
Or how to get two segfaults for the price of one.
In one window start:
Mserver --dbinit 'module(sql_server);
mapi_register(sql_frontend()); mapi_start();'
In another window start:
MapiClient -lsql
and then watch them both get a segfault.
One problem is that the SQL server forgets to do some
important initialization if started through the new
mapi_register method.
## Comment 5100
Date: 2006-03-07 11:48:58 +0100
From: @njnes
Logged In: YES
user_id=43556
The sql_frontend() now creates the sql catalog data
structures. TODO add a test script.
## Comment 5101
Date: 2006-03-08 14:47:58 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Added test src/test/BugTracker/connect_crash.SF-1436626.py.
## Comment 5102
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1436626 at http://sourceforge.net/support/tracker.php?aid=1436626
| SQL connect causes crash | https://api.github.com/repos/MonetDB/MonetDB/issues/884/comments | 0 | 2020-11-30T09:04:14Z | 2024-06-27T11:09:26Z | https://github.com/MonetDB/MonetDB/issues/884 | 753,292,695 | 884 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-21 09:58:40 +0100
From: @drstmane
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-02-21 03:51:35 +0100
## Comment 5095
Date: 2006-02-21 09:58:40 +0100
From: @drstmane
Since yesterday's (Mon, Feb 20 2006) changes in
pathfinder/compiler/mil/milprint_summer.c,1.163.2.3
(boncz):
========
[ 1423349 ] PF: err when "xs:double" is used
bug in evaluateOpOpt (thanks Jennie for figuring that out)
========
or in
MonetDB/src/modules/plain/constant.mx,1.7.2.2 (sjoerd):
========
Added a version of CTrefine for oid,any parameters.
Also changed the names of the formal parameters of a
couple of
functions to be different (i.e. "a" and "b" instead of
both "b").
========
several pathfinder tests fail with:
========
ERROR: interpret: no matching MIL operator to
'reverse(oid)'.
MAYBE YOU MEAN:
reverse(BAT[any::1,any::2]) : BAT[any::2,any::1]
ERROR: interpret_params: {min}(param 1): evaluation error.
ERROR: interpret_params: tmark(param 1): evaluation error.
========
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_BugDay_2005-12-19_0.9.3/not_constant_2.SF-1378750.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music/limit_04.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_mod_MIL/full_04.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_mod_MIL/limit_04.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_mod_XQ/full_04.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_mod_XQ/limit_04.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_opt_MIL/full_02.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_opt_MIL/full_04.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_opt_MIL/full_05.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_opt_MIL/limit_02.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_opt_MIL/limit_04.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_opt_MIL/limit_05.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_opt_XQ/full_02.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_opt_XQ/full_04.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_opt_XQ/full_05.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_opt_XQ/limit_02.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_opt_XQ/limit_04.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_opt_XQ/limit_05.err.00.html
## Comment 5096
Date: 2006-02-21 10:54:21 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
In the generated code for the xquery
doc("not_constant_2.SF-1378750.xml")//volume union
doc("not_constant_2.SF-1378750.xml")//data[not(.//volume)]
there is a line
ipik := sorting.reverse().{min}().tmark(0@0);
It's this line where it goes wrong. In this case, the value
of sorting is 1@0 which it got in the "then" branch of the
if statment just before where there are a bunch of CTrefine
calls (so it might be my change that triggered this).
I don't know what the outcome should be.
Peter?
## Comment 5097
Date: 2006-02-21 15:51:35 +0100
From: @peterboncz
Logged In: YES
user_id=591107
bug was exposed because CTrefine now maintains constants;
a materialize was missing in miltprint_summer.
in the sorting code I actually removed the need for
constant materialization altogether, using a convenience
proc tmark_unique(col,ipik) that for all unique tail
vlaues of a column gets a head value. These head values
end up in the tail of the result, the head is a oid-
sequence starting at 0@0.
tmark_unique(col) = tmark(reverse(kunique(col)),0@0)
## Comment 5098
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1435740 at http://sourceforge.net/support/tracker.php?aid=1435740
| PF: tests fail, complaining about a missing 'reverse(oid)'. | https://api.github.com/repos/MonetDB/MonetDB/issues/883/comments | 0 | 2020-11-30T09:04:11Z | 2024-06-27T11:09:25Z | https://github.com/MonetDB/MonetDB/issues/883 | 753,292,663 | 883 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-18 05:32:31 +0100
From: @drstmane
To: Stefan Manegold <<Stefan.Manegold>>
Version: -- development
Last updated: 2006-02-18 07:09:59 +0100
## Comment 5092
Date: 2006-02-18 17:32:31 +0100
From: @drstmane
Many Pathfinder test fail as debugging output seems to
have "sneaked in" with the latest code changes;
see e.g.:
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/runtime/descendant.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/runtime/descendant-or-self.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/runtime/ancestor.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/runtime/ancestor-or-self.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/runtime/child.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/runtime/parent.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/runtime/following.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/runtime/preceding.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/runtime/following-sibling.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/runtime/preceding-sibling.out.00.html
## Comment 5093
Date: 2006-02-18 19:09:59 +0100
From: @drstmane
Logged In: YES
user_id=572415
fixed in CVS by removing the left-over debugging prints
## Comment 5094
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1434194 at http://sourceforge.net/support/tracker.php?aid=1434194
| PF: tests fail as output is "spoiled" by debugging output | https://api.github.com/repos/MonetDB/MonetDB/issues/882/comments | 0 | 2020-11-30T09:04:08Z | 2024-06-27T11:09:24Z | https://github.com/MonetDB/MonetDB/issues/882 | 753,292,629 | 882 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-18 01:35:26 +0100
From: @drstmane
To: thijs <<thijswes>>
Version: -- development
Last updated: 2006-02-21 09:31:55 +0100
## Comment 5089
Date: 2006-02-18 13:35:26 +0100
From: @drstmane
Since Thijs' yesterday's sjoin.mx changes, test
src/Mxml/Tests/VOCleftpages.milS fails, apparently
returning some odd extra character in its output:
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/xml/.mTests103/GNU.64.64.d-Fedora4/src_Mxml/VOCleftpages.out.00.html
## Comment 5090
Date: 2006-02-21 09:31:24 +0100
From: @drstmane
Logged In: YES
user_id=572415
Fixed by Thijs.
Thanks.
## Comment 5091
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1434125 at http://sourceforge.net/support/tracker.php?aid=1434125
| XML: test src/Mxml/Tests/VOCleftpages.milS fails | https://api.github.com/repos/MonetDB/MonetDB/issues/881/comments | 0 | 2020-11-30T09:04:04Z | 2024-06-27T11:09:23Z | https://github.com/MonetDB/MonetDB/issues/881 | 753,292,580 | 881 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-18 01:26:16 +0100
From: @drstmane
To: thijs <<thijswes>>
Version: -- development
Last updated: 2006-02-21 09:31:00 +0100
## Comment 5086
Date: 2006-02-18 13:26:16 +0100
From: @drstmane
icc -no-gcc -DHAVE_CONFIG_H -I.
-I/export/scratch0/monet/monet.ntv.64.64.d.7916/xml/src/XMLRegions
-I../..
-I/export/scratch0/monet/monet.ntv.64.64.d.7916/xml/src/XMLRegions
-I/export/scratch0/monet/monet.ntv.64.64.d.7916/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB
-I/export/scratch0/monet/monet.ntv.64.64.d.7916/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB/common
-I/export/scratch0/monet/monet.ntv.64.64.d.7916/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB/gdk
-I/export/scratch0/monet/monet.ntv.64.64.d.7916/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB/monet
-I/export/scratch0/monet/monet.ntv.64.64.d.7916/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB/plain
-I/export/scratch0/monet/monet.ntv.64.64.d.7916/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB/contrib
-I/export/scratch0/monet/monet.ntv.64.64.d.7916/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB/mapi
-I/export/scratch0/monet/monet.ntv.64.64.d.7916/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB/C
-DLIBSJOIN -we140 -wd1418 -c99 -Wall -w2 -mp1 -O3
-restrict -unroll -tpp7 -axWP -we266 -Werror
-wd1418,1419,279,310,981,810,444,193,111,177,171,181,764,269,108,188,1357,102,70,1572
-I/usr/include/libxml2 -c sjoin.c -KPIC -DPIC -o
.libs/lib_sjoin_la-sjoin.o
/export/scratch0/monet/monet.ntv.64.64.d.7916/xml/src/XMLRegions/sjoin.mx(307):
error 593: variable "SAo" was set but never used
int SAs,SAe,SAo,SDs, SDo;
^
/export/scratch0/monet/monet.ntv.64.64.d.7916/xml/src/XMLRegions/sjoin.mx(307):
error 593: variable "SDo" was set but never used
int SAs,SAe,SAo,SDs, SDo;
^
/export/scratch0/monet/monet.ntv.64.64.d.7916/xml/src/XMLRegions/sjoin.mx(411):
error 593: variable "SAo" was set but never used
int SAs,SAe,SAo,SDs, SDo;
^
/export/scratch0/monet/monet.ntv.64.64.d.7916/xml/src/XMLRegions/sjoin.mx(411):
error 593: variable "SDo" was set but never used
int SAs,SAe,SAo,SDs, SDo;
^
/export/scratch0/monet/monet.ntv.64.64.d.7916/xml/src/XMLRegions/sjoin.mx(307):
error 593: variable "SAo" was set but never used
int SAs,SAe,SAo,SDs, SDo;
^
/export/scratch0/monet/monet.ntv.64.64.d.7916/xml/src/XMLRegions/sjoin.mx(307):
error 593: variable "SDo" was set but never used
int SAs,SAe,SAo,SDs, SDo;
^
/export/scratch0/monet/monet.ntv.64.64.d.7916/xml/src/XMLRegions/sjoin.mx(411):
error 593: variable "SAo" was set but never used
int SAs,SAe,SAo,SDs, SDo;
^
/export/scratch0/monet/monet.ntv.64.64.d.7916/xml/src/XMLRegions/sjoin.mx(411):
error 593: variable "SDo" was set but never used
int SAs,SAe,SAo,SDs, SDo;
^
compilation aborted for sjoin.c (code 2)
make[4]: *** [lib_sjoin_la-sjoin.lo] Error 1
see also
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/xml/.ntv.64.64.d-Fedora4/make.out
## Comment 5087
Date: 2006-02-21 09:31:00 +0100
From: @drstmane
Logged In: YES
user_id=572415
Fixed by Thijs.
Thanks!
## Comment 5088
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1434120 at http://sourceforge.net/support/tracker.php?aid=1434120
| XML: sjoin.mx fails to compile with icc | https://api.github.com/repos/MonetDB/MonetDB/issues/880/comments | 0 | 2020-11-30T09:04:01Z | 2024-06-27T11:09:22Z | https://github.com/MonetDB/MonetDB/issues/880 | 753,292,536 | 880 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-18 11:56:04 +0100
From: @drstmane
To: Stefan Manegold <<Stefan.Manegold>>
Version: -- development
Last updated: 2006-02-19 04:42:30 +0100
## Comment 5083
Date: 2006-02-18 11:56:04 +0100
From: @drstmane
Since Peter's yesterday's changes/bug-fixes on the
XQuery_0-10 release branchs,
shred_doc issues
========
!WARNING: Cannot reverse order on void columns.
!WARNING: Please materialize them to oid, first.
========
with the following tests:
tests/BugDay_2005-12-19_0.9.3/Tests/file_locked_after_shredding.SF-1238352.MAPI.py
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_BugDay_2005-12-19_0.9.3/file_locked_after_shredding.SF-1238352.err.00.html
tests/BugDay_2005-12-19_0.9.3/Tests/nat_oid_conv.SF-1261765.milS
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_BugDay_2005-12-19_0.9.3/nat_oid_conv.SF-1261765.out.00.html
tests/BugDay_2005-12-19_0.9.3/Tests/type_error_in_function_resolution.SF-1274903.xq
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_BugDay_2005-12-19_0.9.3/type_error_in_function_resolution.SF-1274903.out.00.html
tests/BugDay_2005-12-19_0.9.3/Tests/type_error_in_cast.SF-1281176.xq
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_BugDay_2005-12-19_0.9.3/type_error_in_cast.SF-1281176.out.00.html
## Comment 5084
Date: 2006-02-19 16:42:29 +0100
From: @drstmane
Logged In: YES
user_id=572415
Fixed in MonetDB_4-10 release branch:
instead of issuing a warning, BATorder and BATrevert now
materialize
void columns in-place where necessary (just like other
in-place update operations)
## Comment 5085
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1434084 at http://sourceforge.net/support/tracker.php?aid=1434084
| PF: shred_doc: "Cannot reverse order on void columns." | https://api.github.com/repos/MonetDB/MonetDB/issues/879/comments | 0 | 2020-11-30T09:03:57Z | 2024-06-27T11:09:21Z | https://github.com/MonetDB/MonetDB/issues/879 | 753,292,485 | 879 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-18 11:48:24 +0100
From: @drstmane
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-02-20 12:54:57 +0100
## Comment 5078
Date: 2006-02-18 11:48:24 +0100
From: @drstmane
Since Peter's yesterday's changes/bug-fixes in the
MonetDB_4-10 release branch,
test src/modules/plain/Tests/xatbles.milM fails with
========
!ERROR: CTrefine: both BATs must have the same
cardinality and their heads must form a 1-1 match.
!ERROR: CTrefine: operation failed.
========
(http://monetdb.cwi.nl/testing/projects/monetdb/Stable/MonetDB/.mTests103/GNU.64.64.d-Fedora4/src_modules_plain/xtables.out.00.html)
and test tests/BugReports/Tests/no.148.milS also shows
differences.
## Comment 5079
Date: 2006-02-18 11:49:23 +0100
From: @drstmane
Logged In: YES
user_id=572415
forgot TestWeb URL for second test:
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/MonetDB/.mTests103/GNU.64.64.d-Fedora4/tests_BugReports/no.148.out.00.html
## Comment 5080
Date: 2006-02-18 14:19:04 +0100
From: @drstmane
Logged In: YES
user_id=572415
The same problem occurs with the propagated bug-fixes in the
development trunk:
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/src_modules_plain/xtables.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/tests_BugReports/no.148.out.00.html
## Comment 5081
Date: 2006-02-20 00:54:57 +0100
From: @peterboncz
Logged In: YES
user_id=591107
bugfix in GDKvmalloc,gdkutils.mx
the last 16 bytes of a heap that just borders on a 64KB
boundary where lost.
this bug has been in place for a long time..
it was exposed by setting the VM allocation threshold lower,
and the xtables
scrips allocating bats of nice powers of two (which since
the disappearance of
the BUN hole translate in heaps of nice powers of two).
let's say thank you to regression testing and those who
support it (StM)
## Comment 5082
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1434079 at http://sourceforge.net/support/tracker.php?aid=1434079
| CTrefine failes in xtables tests | https://api.github.com/repos/MonetDB/MonetDB/issues/878/comments | 0 | 2020-11-30T09:03:54Z | 2024-06-27T11:09:20Z | https://github.com/MonetDB/MonetDB/issues/878 | 753,292,444 | 878 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-17 11:56:53 +0100
From: @peterboncz
To: @sjoerdmullender
Version: -- development
Last updated: 2006-02-19 05:13:44 +0100
## Comment 5071
Date: 2006-02-17 23:56:53 +0100
From: @peterboncz
make[9]: Entering directory
`/data/home/boncz/MonetDB/Linux-gcc-
debug/src/mapi/clients/java'
/usr/local/bin/ant -
f ../../../../../src/mapi/clients/java/build.xml -
Dbuilddir=/data/home/boncz/MonetDB/Linux-gcc-
debug/src/mapi/clients/java -
Djardir=/data/home/boncz/MonetDB/Linux-gcc-
debug/src/mapi/clients/java jar_mapi
Unknown argument: -cp
ant [options] [target [target2 [target3] ...]]
Options:
-help, -h print this message
-projecthelp, -p print project help information
-version print the version information
and exit
-diagnostics print information that might
be helpful to
diagnose or report problems.
-quiet, -q be extra quiet
-verbose, -v be extra verbose
-debug, -d print debugging information
-emacs, -e produce logging information
without adornments
-lib <path> specifies a path to search
for jars and classes
-logfile <file> use given file for log
-l <file> ''
-logger <classname> the class which is to perform
logging
-listener <classname> add an instance of class as a
project listener
-noinput do not allow interactive input
-buildfile <file> use given buildfile
-file <file> ''
-f <file> ''
-D<property>=<value> use value for given property
-keep-going, -k execute all targets that do
not depend
on failed target(s)
-propertyfile <name> load all properties from file
with -D
properties taking precedence
-inputhandler <class> the class which will handle
input requests
-find <file> (s)earch for buildfile
towards the root of
-s <file> the filesystem and use it
-nice number A niceness value for the main
thread:
1 (lowest) to 10 (highest); 5
is the default
make[9]: *** [jar_mapi_ant_target] Error 1
make[9]: Leaving directory
`/data/home/boncz/MonetDB/Linux-gcc-
debug/src/mapi/clients/java'
make[8]: *** [install] Error 2
make[8]: Leaving directory
`/data/home/boncz/MonetDB/Linux-gcc-
debug/src/mapi/clients/java'
make[7]: *** [install-recursive] Error 1
make[7]: Leaving directory
`/data/home/boncz/MonetDB/Linux-gcc-
debug/src/mapi/clients'
make[6]: *** [install] Error 2
make[6]: Leaving directory
`/data/home/boncz/MonetDB/Linux-gcc-
debug/src/mapi/clients'
make[5]: *** [install-recursive] Error 1
make[5]: Leaving directory
`/data/home/boncz/MonetDB/Linux-gcc-debug/src/mapi'
make[4]: *** [install] Error 2
make[4]: Leaving directory
`/data/home/boncz/MonetDB/Linux-gcc-debug/src/mapi'
make[3]: *** [install-recursive] Error 1
make[3]: Leaving directory
`/data/home/boncz/MonetDB/Linux-gcc-debug/src'
make[2]: *** [install] Error 2
make[2]: Leaving directory
`/data/home/boncz/MonetDB/Linux-gcc-debug/src'
make[1]: *** [install-recursive] Error 1
make[1]: Leaving directory
`/data/home/boncz/MonetDB/Linux-gcc-debug'
make: *** [install] Error 2
## Comment 5072
Date: 2006-02-18 00:14:48 +0100
From: @drstmane
Logged In: YES
user_id=572415
The problem seems to be that /usr/local/bin/ant is broken
while /usr/bin/ant seems to work fine:
========
00:08:47 manegold@sn-1:~ $ /usr/bin/ant -version
Apache Ant version 1.6.2 compiled on May 23 2005
--------
00:08:52 manegold@sn-1:~ $ /usr/local/bin/ant -version
Unknown argument: -cp
ant [options] [target [target2 [target3] ...]]
Options:
-help, -h print this message
-projecthelp, -p print project help information
-version print the version information and exit
-diagnostics print information that might be
helpful to
diagnose or report problems.
-quiet, -q be extra quiet
-verbose, -v be extra verbose
-debug, -d print debugging information
-emacs, -e produce logging information without
adornments
-lib <path> specifies a path to search for jars
and classes
-logfile <file> use given file for log
-l <file> ''
-logger <classname> the class which is to perform logging
-listener <classname> add an instance of class as a
project listener
-noinput do not allow interactive input
-buildfile <file> use given buildfile
-file <file> ''
-f <file> ''
-D<property>=<value> use value for given property
-keep-going, -k execute all targets that do not depend
on failed target(s)
-propertyfile <name> load all properties from file with -D
properties taking precedence
-inputhandler <class> the class which will handle input
requests
-find <file> (s)earch for buildfile towards the
root of
-s <file> the filesystem and use it
-nice number A niceness value for the main thread:
1 (lowest) to 10 (highest); 5 is
the default
========
does anyone require the ant 1.6.5 in /usr/local/bin/ ?
or is the ant 1.6.2 in /usr/bin/ enough?
if the latter, we should consider removing the ant in
/usr/local/bin/
Till then, the work around is to configure with
`--with-ant=/usr/bin/ant`.
## Comment 5073
Date: 2006-02-18 02:29:03 +0100
From: @peterboncz
Logged In: YES
user_id=591107
right, thanks Stefan.
so now I configure with --with-ant=/usr/bin/ant
and indeed it gets further.. almost to the end.
now it is the Mbedded compilation in src/tools that fails.
cd /data/home/boncz/rel/MonetDB/Linux-gcc-
debug/src/tools; /bin/sh ../../libtool --tag=CC --
mode=relink gcc -m64 -std=c99 -Wall -W -g -Werror-implicit-
function-declaration -Werror -Wno-unused-function -Wno-
unused-label -D_REENTRANT -o libembeddedmil.la -
rpath /data/home/boncz/rel/MonetDB/Linux-gcc-debug/lib
libembeddedmil_la-embeddedclient.lo
libinitmodules.la ../mapi/clients/C/libMapi.la ../monet/libm
onet.la ../modules/plain/lib_algebra.la ../modules/plain/lib
_arith.la ../modules/plain/lib_bat.la ../modules/plain/lib_b
uiltin.la ../modules/plain/lib_str.la ../modules/plain/lib_s
ys.la ../modules/plain/lib_trans.la ../modules/plain/lib_con
stant.la ../gdk/libbat.la -ldl -
lpthread ../common/libmutils.la ../common/libstream.la )
mv: cannot stat `libembeddedmil.so.0.0.0': No such file or
directory
libtool: install: error: relink `libembeddedmil.la' with
the above command before installing it
make[6]: *** [install-embeddedmilLTLIBRARIES] Error 1
make[6]: Leaving directory
`/data/home/boncz/rel/MonetDB/Linux-gcc-debug/src/tools'
make[5]: *** [install-am] Error 2
make[5]: Leaving directory
`/data/home/boncz/rel/MonetDB/Linux-gcc-debug/src/tools'
make[4]: *** [install] Error 2
make[4]: Leaving directory
`/data/home/boncz/rel/MonetDB/Linux-gcc-debug/src/tools'
make[3]: *** [install-recursive] Error 1
make[3]: Leaving directory
`/data/home/boncz/rel/MonetDB/Linux-gcc-debug/src'
make[2]: *** [install] Error 2
make[2]: Leaving directory
`/data/home/boncz/rel/MonetDB/Linux-gcc-debug/src'
make[1]: *** [install-recursive] Error 1
make[1]: Leaving directory
`/data/home/boncz/rel/MonetDB/Linux-gcc-debug'
make: *** [install] Error 2
## Comment 5074
Date: 2006-02-18 09:23:45 +0100
From: @njnes
Logged In: YES
user_id=43556
The ant is fixed (changed -cp into -lib in the call to java).
I don't get the embedded problems.
## Comment 5075
Date: 2006-02-18 23:49:24 +0100
From: @drstmane
Logged In: YES
user_id=572415
Also for me, compilation of both MonetDB & Pathfinder
(latest CVS heads) (now) works out-of-the-box without any
problems on sn-1.
Peter,
do you still have problems, or can we close this bug report?
## Comment 5076
Date: 2006-02-19 17:13:44 +0100
From: @peterboncz
Logged In: YES
user_id=591107
is now resolved.
probably my mistake, or the result of running make in a
subdirectory, that failed first, and when the ant problem
was gone left over stuff that impeded it to succeed later
## Comment 5077
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1433887 at http://sourceforge.net/support/tracker.php?aid=1433887
| MonetDB clean 'make install' fails on sn-1 | https://api.github.com/repos/MonetDB/MonetDB/issues/877/comments | 0 | 2020-11-30T09:03:50Z | 2024-06-27T11:09:19Z | https://github.com/MonetDB/MonetDB/issues/877 | 753,292,393 | 877 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-17 12:52:39 +0100
From: @drstmane
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-02-19 05:45:42 +0100
## Comment 5066
Date: 2006-02-17 12:52:39 +0100
From: @drstmane
Since Peter's yesterday's changes, shred_doc() seems to
fail with
!FATAL: BATpropcheck: BAT tmp_107 has inconsistent
descriptor 1
see:
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_BugTracker/auctionG_xml.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_BugsViaSourgeforce/auctionG_xml.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music/import_doc.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_mod_MIL/import_doc.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_mod_XQ/import_doc.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_opt_MIL/import_doc.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_Music_opt_XQ/import_doc.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_WebSite/voc.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XMark/import_doc.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XMark_alg/import_doc.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XMark_mod_MIL/import_doc.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XMark_mod_XQ/import_doc.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XMark_opt_MIL/import_doc.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XMark_opt_XQ/import_doc.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XPath/import_doc.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XPathMark/document_1.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XQuery/import_doc.err.00.html
## Comment 5067
Date: 2006-02-19 16:51:44 +0100
From: @drstmane
Logged In: YES
user_id=572415
The error message seems to be triggerd by
((b->batBuns->free > b->batBuns->size) << 12)
in MonetDB/src/gdk/gdk_bat.mx's BATpropcheck().
## Comment 5068
Date: 2006-02-19 17:45:20 +0100
From: @peterboncz
Logged In: YES
user_id=591107
this bug only occurs with debugmask 8; that's why I never
hit it.
The fatal is given in BATprocheck, it complains if
batBuns->free > batBuns->size; on the pre_nid bat (since
recently of the void,void kind)
however, I was instructed to do exactly this (making free >
size) by Niels (none other than the father of the void,void
bat).
strange that this never ocurred before.
So I had a beautiful fix: shoot the messenger (i.e. remove
this check from BATpropcheck).
Acoompanied by a small optimization: instead of BATTINY
(1024 entries) we allocate only one BUN for void,void bats.
## Comment 5069
Date: 2006-02-19 18:09:36 +0100
From: @drstmane
Logged In: YES
user_id=572415
Peter,
please don't forget to check-in your fixes to MonetDB
;-)
## Comment 5070
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1433507 at http://sourceforge.net/support/tracker.php?aid=1433507
| PF: shred_doc fails with BATproperty error | https://api.github.com/repos/MonetDB/MonetDB/issues/876/comments | 0 | 2020-11-30T09:03:46Z | 2024-06-27T11:09:18Z | https://github.com/MonetDB/MonetDB/issues/876 | 753,292,346 | 876 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-17 12:41:10 +0100
From: @drstmane
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-02-19 05:01:31 +0100
## Comment 5063
Date: 2006-02-17 12:41:10 +0100
From: @drstmane
With Peter's yesterdays changes to src/gdk/gdk.mx
("'optimized' / 'more cautious' version of BUNfndVOID")
the semantics of find(void(nil)) on a void-nil headed
BAT seem to have changed:
Before, the find succeeded, now it fails;
see tests
tests/BugDay_2005-10-06_4.9.3/Tests/find_nil_in_void.SF-921814.milS
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/tests_BugDay_2005-10-06_4.9.3/find_nil_in_void.SF-921814.out.00.html
and
tests/BugsViaSourgeforce/Tests/ID.921814.milS
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/tests_BugsViaSourgeforce/ID.921814.out.00.html
for details.
If these semantic changes were intended, the respective
stable output needs to be approved accordingly.
Otherwise, the respective code needs to be fixed to
restore the original semantic.
## Comment 5064
Date: 2006-02-19 17:01:06 +0100
From: @peterboncz
Logged In: YES
user_id=591107
ok, I modified my re-implementation such that find(nil) of a
void column returns the first BUN.
it remains strange semantics.. but I do not care enough to
carry through such a change.
## Comment 5065
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1433498 at http://sourceforge.net/support/tracker.php?aid=1433498
| semantics of find(BAT[void-nil,any], void(nil)) has changed | https://api.github.com/repos/MonetDB/MonetDB/issues/875/comments | 0 | 2020-11-30T09:03:43Z | 2024-06-27T11:09:17Z | https://github.com/MonetDB/MonetDB/issues/875 | 753,292,310 | 875 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-17 12:21:57 +0100
From: @drstmane
To: @yzchang
Version: -- development
Last updated: 2006-02-21 03:08:46 +0100
## Comment 5058
Date: 2006-02-17 12:21:57 +0100
From: @drstmane
icc -no-gcc -DHAVE_CONFIG_H -I.
-I/export/scratch0/monet/monet.ntv.64.64.d.32666/pathfinder/runtime
-I..
-I/export/scratch0/monet/monet.ntv.64.64.d.32666/pathfinder/runtime
-I/export/scratch0/monet/monet.ntv.64.64.d.32666/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB
-I/export/scratch0/monet/monet.ntv.64.64.d.32666/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB/common
-I/export/scratch0/monet/monet.ntv.64.64.d.32666/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB/gdk
-I/export/scratch0/monet/monet.ntv.64.64.d.32666/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB/monet
-I/export/scratch0/monet/monet.ntv.64.64.d.32666/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB/plain
-I/export/scratch0/monet/monet.ntv.64.64.d.32666/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB/contrib
-I/export/scratch0/monet/monet.ntv.64.64.d.32666/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB/mapi
-I/export/scratch0/monet/monet.ntv.64.64.d.32666/MonetDB/.ntv.64.64.d-Fedora4/include/MonetDB/C
-I/usr/include/libxml2 -I../compiler/include
-I/export/scratch0/monet/monet.ntv.64.64.d.32666/pathfinder/runtime/../compiler/include
-DLIBPF_SOAP -we140 -wd1418 -c99 -Wall -w2 -mp1 -O3
-restrict -unroll -tpp7 -axWP -we266 -Werror
-wd1418,1419,279,310,981,810,444,193,111,177,171,181,764,269,108,188,1357,102,70,1572
-DEMBEDDED -DMT -DNO_CGI -DNO_AUTH -DIO_MAX=10240 -c
pf_soap.c -KPIC -DPIC -o .libs/lib_pf_soap_la-pf_soap.o
/export/scratch0/monet/monet.ntv.64.64.d.32666/pathfinder/runtime/pf_soap.mx(460):
error 593: variable "level" was set but never used
char *strptr = NULL, *kind = NULL, *level = NULL,
*text_base = NULL;
^
compilation aborted for pf_soap.c (code 2)
make[2]: *** [lib_pf_soap_la-pf_soap.lo] Error 1
see also
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.ntv.64.64.d-Fedora4/make.out
## Comment 5059
Date: 2006-02-21 15:08:45 +0100
From: @yzchang
Logged In: YES
user_id=341633
Fixed in CVS head: remove the unused variable.
## Comment 5060
Date: 2006-03-02 15:00:45 +0100
From: @yzchang
Logged In: YES
user_id=341633
Fixed in CVS head: remove the unused variable.
## Comment 5061
Date: 2006-03-02 16:14:25 +0100
From: @yzchang
Logged In: YES
user_id=341633
Close this bug.
## Comment 5062
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1433488 at http://sourceforge.net/support/tracker.php?aid=1433488
| PF: runtime/pf_soap.mx failes to compile with icc | https://api.github.com/repos/MonetDB/MonetDB/issues/874/comments | 0 | 2020-11-30T09:03:39Z | 2024-06-27T11:09:16Z | https://github.com/MonetDB/MonetDB/issues/874 | 753,292,271 | 874 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-17 01:06:14 +0100
From: @drstmane
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-02-17 09:16:29 +0100
## Comment 5055
Date: 2006-02-17 01:06:14 +0100
From: @drstmane
The latest version of runtime/shredder.mx (r1.64) fails
to compile with
gcc -m64 -DHAVE_CONFIG_H -I.
-I/ufs/manegold/_/scratch0/Monet/Testing/Current/pathfinder/runtime
-I..
-I/ufs/manegold/_/scratch0/Monet/Testing/Current/pathfinder/runtime
-I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d.--enable-debug.PREFIX./MONETDB/include/MonetDB
-I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d.--enable-debug.PREFIX./MONETDB/include/MonetDB/common
-I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d.--enable-debug.PREFIX./MONETDB/include/MonetDB/gdk
-I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d.--enable-debug.PREFIX./MONETDB/include/MonetDB/monet
-I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d.--enable-debug.PREFIX./MONETDB/include/MonetDB/plain
-I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d.--enable-debug.PREFIX./MONETDB/include/MonetDB/contrib
-I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d.--enable-debug.PREFIX./MONETDB/include/MonetDB/mapi
-I/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d.--enable-debug.PREFIX./MONETDB/include/MonetDB/C
-I/usr/include/libxml2 -I../compiler/include
-I/ufs/manegold/_/scratch0/Monet/Testing/Current/pathfinder/runtime/../compiler/include
-DLIBPF_SUPPORT -std=c99 -Wall -W -g
-Werror-implicit-function-declaration -Werror
-Wno-unused-function -Wno-unused-label -DEMBEDDED -DMT
-DNO_CGI -DNO_AUTH -DIO_MAX=10240 -c shredder.c -fPIC
-DPIC -o .libs/lib_pf_support_la-shredder.o
/ufs/manegold/_/scratch0/Monet/Testing/Current/pathfinder/runtime/shredder.mx:
In function 'CMDshred2bats':
/ufs/manegold/_/scratch0/Monet/Testing/Current/pathfinder/runtime/shredder.mx:647:
error: 'REMAP_PAGE_SIZE' undeclared (first use in this
function)
/ufs/manegold/_/scratch0/Monet/Testing/Current/pathfinder/runtime/shredder.mx:647:
error: (Each undeclared identifier is reported only once
/ufs/manegold/_/scratch0/Monet/Testing/Current/pathfinder/runtime/shredder.mx:647:
error: for each function it appears in.)
make[2]: *** [lib_pf_support_la-shredder.lo] Error 1
make[2]: Leaving directory
`/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d.--enable-debug.BUILD./PATHFINDER/runtime'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory
`/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/.GNU.64.64.d.--enable-debug.BUILD./PATHFINDER'
make: *** [all] Error 2
## Comment 5056
Date: 2006-02-17 09:16:29 +0100
From: @drstmane
Logged In: YES
user_id=572415
Oops, my fault: forgot to re-compile MonetDB, first. Sorry!
## Comment 5057
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1433263 at http://sourceforge.net/support/tracker.php?aid=1433263
| PF: runtime/shredder.mx r1.64 fails to compile | https://api.github.com/repos/MonetDB/MonetDB/issues/873/comments | 0 | 2020-11-30T09:03:37Z | 2024-06-28T13:34:15Z | https://github.com/MonetDB/MonetDB/issues/873 | 753,292,242 | 873 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-16 03:50:50 +0100
From: Christian Gruen <<christiangruen>>
To: XQuery devs <<bugs-xquery>>
Version: -- development
Duplicates: #871
Last updated: 2010-05-04 11:02:38 +0200
## Comment 5053
Date: 2006-02-16 15:50:50 +0100
From: Christian Gruen <<christiangruen>>
Hi there,
I'm running MonetDB/XQuery 0.10.0 on a 64Bit machine
(Dual 2.2 Opteron, Suse 10.0, 16 GB RAM).
I've downloaded the XML version of the Wikipedia (http:
//download.wikimedia.org/enwiki/20060208/ - pages-
articles.xml.bz2) which occupies 5 GB as unzipped file.
I could successfully shred it into MonetDB, but when I
send XPath queries, the behaviour is pretty
unpredictable.
I could successfully process the following example
queries (but sometimes the Mserver didn't react
anymore, and the memory consumption climbed up to 15
GB):
- doc("wikipedia.xml")//page[title = 'Konstanz'] (31
Sec.)
- doc("wikipedia.xml")//title[. = 'Konstanz']/parent::
page
The upcoming queries didn't yield any results at all:
- doc("wikipedia.xml")/mediawiki/page/title[text() =
'Konstanz']/..
- doc("wikipedia.xml")//page/title[. = 'Konstanz']/..
..the Monet server leaves within milliseconds, but does
not throw any errors or warnings.
I've successfully shredded XMark instances with 1 and
10 GB. I assumed that the node result sets might get
too big - but the Wikipedia mainly has full text
entries and distinctly less tree nodes than the XMark
documents. Do you have a clue?
Regards,
Christian Gruen
## Comment 5054
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1432923 at http://sourceforge.net/support/tracker.php?aid=1432923
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-common@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
Bug has invalid status, setting status to "NEW".
Previous status was "DELETED".
## Comment 13970
Date: 2010-05-04 11:02:38 +0200
From: @grobian
*** This bug has been marked as a duplicate of bug #871 ***
| MonetDB/XQuery & 5 GB XML File | https://api.github.com/repos/MonetDB/MonetDB/issues/872/comments | 0 | 2020-11-30T09:03:33Z | 2024-06-28T07:36:50Z | https://github.com/MonetDB/MonetDB/issues/872 | 753,292,187 | 872 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-16 03:04:00 +0100
From: Christian Gruen <<christiangruen>>
To: GDK devs <<bugs-common>>
Version: -- development
Last updated: 2010-05-04 11:02:38 +0200
## Comment 5048
Date: 2006-02-16 15:04:00 +0100
From: Christian Gruen <<christiangruen>>
Hi there,
I'm running MonetDB/XQuery 0.10.0 on a 64Bit machine
(Dual 2.2 Opteron, Suse 10.0, 16 GB RAM).
I've downloaded the XML version of the Wikipedia (http:
//download.wikimedia.org/enwiki/20060208/ - pages-
articles.xml.bz2) which occupies 5 GB as unzipped file.
I could successfully shred it into MonetDB, but when I
send XPath queries, the behaviour is pretty
unpredictable.
I could successfully process the following example
queries (but sometimes the Mserver didn't react
anymore, and the memory consumption climbed up to 15
GB):
- doc("wikipedia.xml")//page[title = 'Konstanz'] (31
Sec.)
- doc("wikipedia.xml")//title[. = 'Konstanz']/parent::
page
The upcoming queries didn't yield any results at all:
- doc("wikipedia.xml")/mediawiki/page/title[text() =
'Konstanz']/..
- doc("wikipedia.xml")//page/title[. = 'Konstanz']/..
..the Monet server leaves within milliseconds, but does
not throw any errors or warnings.
I've successfully shredded XMark instances with 1 and
10 GB. I assumed that the node result sets might get
too big - but the Wikipedia mainly has full text
entries and distinctly less tree nodes than the XMark
documents. Do you have a clue?
Regards,
Christian Gruen
## Comment 5049
Date: 2006-02-18 01:19:36 +0100
From: @peterboncz
Logged In: YES
user_id=591107
not really, there is a bugfix release upcoming that will
improve shredding performance on 64-bits systems for large
documents like yours.
in general, pathfinder works better if a selection is
expressed in a where clause then when it's expressed in a
condition.
it likes:
for $x in expr where cond return $x
better than
expr[cond]
they seem equivalent, but there are subtle differences in
semantics, and currently one is not rewritten into the
other (and the where-case is better optimized).
I will give your queries a try.
## Comment 5050
Date: 2006-02-18 22:48:40 +0100
From: @peterboncz
Logged In: YES
user_id=591107
Hi Christian,
I tested this and indeed we can't say that MonetDB/XQuery
is doing this job very fast. But I cannot reproduce any
crashes (yet).
I compiled the current state of the 0.10 release branch,
which is not exactly the same as the binary release (as I
told you, we will soon be doing a bugfix release).
I compiled with debugging (no optimization), and with 32-
bits oids, which turned out to be **just** enough for this
collection: it indeed has a lot of text data, which takes
3.7GB (with 32-bits oids text columns are limited to 4GB).
But, it works. Your queries all return empty results
because the wikipedia document defines a namespace and you
don't use that in your queries. Thus, wherever you use an
element name X in your queries, I typed *:X instead. Then
your Konstanz queries produce the expected result.
When I do this, queries 1 and 3 run in about 100 seconds.
Maybe it cooincides with your 30 when I would have compiled
with optimization.
The reason for the slowness is that each query actually
extracts the string values of all title elements. This step
alone takes 70 seconds on my system. Regrettably, //title
is not very selective (2.5M out of all 12.5M elements). It
would be better if we had value-indices on as that support
string and/or decimal queries. This way we could avoid a
linear scan.
Currently, there are only indices on element names. That is
why MonetDB/XQuery is particularly effective
on /descendant:X queries; and child is fast as well. All
other axis are fast as long as they are used outside any
for-loop (or nested in only very vew iterations).
Especially for the often-used parent and ancestor steps,
the problems with nested exeuction are sometimes annoying.
Queries 2 and 4 are even more expensive (200 seconds) have
the problem that you do a string comparison on a node: [.
= 'Konstanz']. This leads to more expensive execution:
reconstructing the text value of an element node (note that
in XML this requires concatenating the string values of all
text node descendants). Thus, I recommend using [text()
= 'Konstanz'] instead.
So, I have a question back: when you say that Mserver
leaves, you mean it crashes? Or you mean MapiClient gave
back an empty result?
Peter
## Comment 5051
Date: 2006-02-20 14:19:02 +0100
From: Christian Gruen <<christiangruen>>
Logged In: YES
user_id=1352644
Hi Peter,
thanks for your extensive response and the details on query
processing. The major problem was in fact that I forgot to
add the namespace.. I get results for all of the queries now.
With 'Mserver leaves' I meant that I got back an empty
result set - which seems clear now as query 3 and 4 start
with the root node, and as I omitted the namespace, parsing
was immediately stopped. In query 1 and 2 all descendant
nodes were parsed for the specified tag, so that's why I
thought (by mistake) that there were some results.
Memory consumption is still climbing pretty high, esp. when
lots of string are extracted (which is what as you already
said), so I'll also give the 32bit version a try.
Do you think that value indexing will be integrated Monet/PF
in near future? I'm currently experimenting with a simple
XMLDB/XPath prototype (pos/parent based) to find out how to
smoothly integrate value indexing and combine it with other
location steps & predicates. Issues such as atomization are
in fact tedious when you try to find efficient solutions here..
Thanks & cheers!
Christian
## Comment 5052
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1432897 at http://sourceforge.net/support/tracker.php?aid=1432897
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-common@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
## Comment 13971
Date: 2010-05-04 11:02:38 +0200
From: @grobian
*** Bug #872 has been marked as a duplicate of this bug. ***
| MonetDB/XQuery & 5 GB XML File | https://api.github.com/repos/MonetDB/MonetDB/issues/871/comments | 0 | 2020-11-30T09:03:30Z | 2024-06-28T07:36:49Z | https://github.com/MonetDB/MonetDB/issues/871 | 753,292,151 | 871 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-15 09:13:12 +0100
From: @grobian
To: @njnes
Version: -- development
Last updated: 2006-02-18 04:40:15 +0100
## Comment 5045
Date: 2006-02-15 21:13:12 +0100
From: @grobian
tapestry-> SELECT "current_schema" FROM "sys"."sessions";
+----------+
| tapestry |
+==========+
| tapestry |
| tapestry |
| tapestry |
| tapestry |
| tapestry |
| tapestry |
| tapestry |
+----------+
7 rows
tapestry-> select current_schema;
+----------+
| tapestry |
+==========+
| tapestry |
+----------+
1 row
tapestry-> select "now";
Error: identifier now unknown or ambigious
tapestry-> select "now()";
+--------------+
| single_value |
+==============+
| now() |
+--------------+
1 row
tapestry-> select "current_schema";
+----------+
| tapestry |
+==========+
| tapestry |
+----------+
1 row
It looks incorrect to me that even a quoted occurence
of current_schema" is interpreted as not an ident.
## Comment 5046
Date: 2006-02-18 16:40:15 +0100
From: @njnes
Logged In: YES
user_id=43556
indeed when an quoted string doesn't exist in the current
scope, ie no identifiers with this name exist, it is taken as
variable, ie if the variable exists in the sys.sessions table
it returns a that value. I consider this a feature, ie
closing this bug report.
## Comment 5047
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1432426 at http://sourceforge.net/support/tracker.php?aid=1432426
| SQL: current_schema taken as function when in quotes | https://api.github.com/repos/MonetDB/MonetDB/issues/870/comments | 0 | 2020-11-30T09:03:27Z | 2024-06-28T07:36:49Z | https://github.com/MonetDB/MonetDB/issues/870 | 753,292,117 | 870 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-15 01:52:40 +0100
From: @grobian
To: clients devs <<bugs-clients>>
Version: -- development
Last updated: 2006-02-15 03:21:42 +0100
## Comment 5042
Date: 2006-02-15 13:52:40 +0100
From: @grobian
On various places in the Mapi library, error messages
are written to stdout. For a library this is a weird
thing to do, unless there is some flag to
enable/disable it (preferable to set a stream to write to).
## Comment 5043
Date: 2006-02-15 15:21:42 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Fixed. No test.
## Comment 5044
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1432137 at http://sourceforge.net/support/tracker.php?aid=1432137
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-clients@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
| MAPI: error messages are written to stdout | https://api.github.com/repos/MonetDB/MonetDB/issues/869/comments | 0 | 2020-11-30T09:03:25Z | 2024-06-27T11:09:11Z | https://github.com/MonetDB/MonetDB/issues/869 | 753,292,088 | 869 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-15 01:50:57 +0100
From: @grobian
To: Fabian Groffen <<fabian>>
Version: -- development
Last updated: 2006-02-22 02:07:00 +0100
## Comment 5037
Date: 2006-02-15 13:50:57 +0100
From: @grobian
mid = mapi_connect(hostname, port, username,
password, language);
if (mapi_error(mid)) {
printf("Error: %s\n", mapi_error_str(mid));
}
In the above example mapi_error_str returns NULL even
if there is an error message from the server. The
message is written to stdout, which is another bug, but
indicates there is a message.
## Comment 5038
Date: 2006-02-15 15:21:03 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Fixed in CVS.
Do you have an idea for a test?
## Comment 5039
Date: 2006-02-15 18:44:01 +0100
From: @grobian
Logged In: YES
user_id=963970
yeah, connect to an SQL database with an invalid
username/password. I don't think doing that through
MapiClient makes much sense, so either we make a php file,
or just extend one of the C samples and let it use wrong
credentials.
## Comment 5040
Date: 2006-02-22 14:07:00 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Now it's really fixed, and I added a test to boot.
sql/src/test/BugTracker/Tests/mapi_connect_errors_vanish.SF-1432134.*
## Comment 5041
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1432134 at http://sourceforge.net/support/tracker.php?aid=1432134
| MAPI: mapi_connect errors vanish | https://api.github.com/repos/MonetDB/MonetDB/issues/868/comments | 0 | 2020-11-30T09:03:21Z | 2024-06-27T11:09:10Z | https://github.com/MonetDB/MonetDB/issues/868 | 753,292,041 | 868 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-14 11:15:45 +0100
From: Wouter Alink <<alink>>
To: @sjoerdmullender
Version: -- development
Last updated: 2006-02-21 02:00:52 +0100
## Comment 5030
Date: 2006-02-14 11:15:45 +0100
From: Wouter Alink <<alink>>
Hopefully this one can go into the bug-fix-release, it
is very annoying:
the query (selecting a non-existent attribute and
asking for distinct-values):
echo 'distinct-values(<a><b aap="test"/></a>//@beer)' |
MapiClient -lxq
produces:
ERROR: interpret: no matching MIL operator to
'CTrefine(oid, int)'.
MAYBE YOU MEAN:
CTrefine(BAT[oid,any], BAT[oid,any]) : BAT[oid,oid]
CTrefine(oid, BAT[oid,any]) : BAT[oid,oid]
CTrefine(BAT[oid,oid], any::1) : BAT[oid,oid]
## Comment 5031
Date: 2006-02-15 17:13:07 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
In the generated MIL code there is a line
sorting := sorting.CTrefine(kind);
It's this line that give the problem. When this line is
executed, the value of sorting is 1@0, and the value of kind
is 34, i.e. an oid and an int. I can fix this by adding a
proc to the constant module in MonetDB:
proc CTrefine ( oid a, any b) : oid
{
return a;
}
With this proc added the query runs and gives an empty
result (i.e. the command returns an empty line). Is that
the desired result?
## Comment 5032
Date: 2006-02-20 11:21:40 +0100
From: @peterboncz
Logged In: YES
user_id=591107
yes sjoerd, that's ok. It is a indeed a constant handling
problem.
I propose to add this to the 0.10 bugfix release as well.
## Comment 5033
Date: 2006-02-20 13:31:54 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Fix is checked in on the stable branch of MonetDB.
## Comment 5034
Date: 2006-02-21 12:58:45 +0100
From: @drstmane
Logged In: YES
user_id=572415
re-opened to remind us that we need to add a test script to CVS.
## Comment 5035
Date: 2006-02-21 14:00:52 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Test added:
tests/BugTracker/Tests/empty_distinct_values.SF-1431357.*
## Comment 5036
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1431357 at http://sourceforge.net/support/tracker.php?aid=1431357
| XQuery: distinct-values on empty list | https://api.github.com/repos/MonetDB/MonetDB/issues/867/comments | 0 | 2020-11-30T09:03:18Z | 2024-06-27T11:09:09Z | https://github.com/MonetDB/MonetDB/issues/867 | 753,291,991 | 867 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-13 09:44:43 +0100
From: @MarcinZukowski
To: @njnes
Version: -- development
Last updated: 2006-02-14 12:28:32 +0100
## Comment 5026
Date: 2006-02-13 21:44:43 +0100
From: @MarcinZukowski
Here's the script:
<<<
PROC print_stats() : void
{
var b := view_gdk_bbp();
b:=nil;
printf("vm_cursize %lld\n", vm_cursize());
}
while(true) {
print_stats();
}
<<<
If you let it run, you'll see the memory growth.
I don't know if it's related to a recent BAT leak bug,
hence I'm submitting as a new one.
## Comment 5027
Date: 2006-02-14 00:28:32 +0100
From: @njnes
Logged In: YES
user_id=43556
fixed, a leak of bat idents
## Comment 5028
Date: 2006-06-20 14:34:29 +0200
From: @njnes
Logged In: YES
user_id=43556
test exits
./tests/BugTracker/Tests/leak_of_view_gdk_bbp.SF-1430968.milS
## Comment 5029
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1430968 at http://sourceforge.net/support/tracker.php?aid=1430968
| view_gdk_bbp leaking | https://api.github.com/repos/MonetDB/MonetDB/issues/866/comments | 0 | 2020-11-30T09:03:14Z | 2024-06-27T11:09:08Z | https://github.com/MonetDB/MonetDB/issues/866 | 753,291,946 | 866 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-13 01:01:16 +0100
From: @grobian
To: @njnes
Version: -- development
Last updated: 2006-02-14 10:32:49 +0100
## Comment 5022
Date: 2006-02-13 13:01:16 +0100
From: @grobian
From our examples on the web:
monetdb-> CREATE USER "voc" WITH PASSWORD 'voc' NAME
'VOC Explorer' SCHEMA "sys";
Operation successful
monetdb-> select * from users;
+---------+---------------+----------------+
| name | fullname | default_schema |
+=========+===============+================+
| monetdb | MonetDB Admin | 1009574 |
| voc | VOC Explorer | 1009574 |
+---------+---------------+----------------+
2 rows
monetdb-> CREATE SCHEMA "voc" AUTHORIZATION "voc";
Operation successful
monetdb-> ALTER USER "voc" SET SCHEMA "voc";
Operation successful
{pegatoo:~} fabian% JdbcClient -p50000 -uvoc
password:
Database connect failed: ERROR: interpret: no matching
MIL operator to 'mvc_bind(void, str, str, str, int)'.
MAYBE YOU MEAN:
mvc_bind(mvc, str, str, str, int) :
BAT[any::1,any::2]
{pegatoo:~} fabian% JdbcClient -p50000 -umonetdb
password:
Database connect failed: ERROR: interpret: no matching
MIL operator to 'mvc_bind(void, str, str, str, int)'.
MAYBE YOU MEAN:
mvc_bind(mvc, str, str, str, int) :
BAT[any::1,any::2]
Even as monetdb user, one cannot login to the server again.
## Comment 5023
Date: 2006-02-14 10:32:49 +0100
From: @njnes
Logged In: YES
user_id=43556
fixed bug . The myc per client global variable was destroyed
in the mvc_sql call (used when sql is rewriten into multi-sql
statements, in this case to insert the user privileges)
## Comment 5024
Date: 2006-06-20 14:32:31 +0200
From: @njnes
Logged In: YES
user_id=43556
test exits
./src/test/BugTracker/Tests/authorization.SF-1430616-data.sql
## Comment 5025
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1430616 at http://sourceforge.net/support/tracker.php?aid=1430616
| SQL: creating a user and schema renders database unusable | https://api.github.com/repos/MonetDB/MonetDB/issues/865/comments | 0 | 2020-11-30T09:03:11Z | 2024-06-27T11:09:07Z | https://github.com/MonetDB/MonetDB/issues/865 | 753,291,909 | 865 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-11 02:16:23 +0100
From: @grobian
To: @njnes
Version: -- development
Last updated: 2006-06-16 02:06:57 +0200
## Comment 5015
Date: 2006-02-11 14:16:23 +0100
From: @grobian
http://monetdb.cwi.nl/TechDocs/APIs/Mapi/C/index.html
is unreadable. It is one large blob of text containing
all kinds of tex-like commands in it. The koala
testing version is the same.
## Comment 5016
Date: 2006-03-09 12:46:29 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Fixed mkdoc.py. Let's wait for testing to see whether this
was enough of a fix.
## Comment 5017
Date: 2006-03-21 14:27:47 +0100
From: @grobian
Logged In: YES
user_id=963970
This looks still as garbled as it was to me.
Bump!
## Comment 5018
Date: 2006-04-14 15:01:42 +0200
From: @sjoerdmullender
Logged In: YES
user_id=43607
The web pages have to be pushed out. It looks like "make
docs" does it right, but the generated pages have to pushed
to the web site.
## Comment 5019
Date: 2006-06-06 21:25:09 +0200
From: @grobian
Logged In: YES
user_id=963970
On koala this still is as much as garbage as it was. How
can the docs be pushed out to the website?
## Comment 5020
Date: 2006-06-11 17:37:55 +0200
From: @mlkersten
Logged In: YES
user_id=490798
Prepared a 'hardwired' version for inclusion in the website.
Once correctly displayed it can be closed
## Comment 5021
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1429688 at http://sourceforge.net/support/tracker.php?aid=1429688
| Website: MAPI documentation is hardly readable | https://api.github.com/repos/MonetDB/MonetDB/issues/864/comments | 0 | 2020-11-30T09:03:08Z | 2024-06-27T11:09:06Z | https://github.com/MonetDB/MonetDB/issues/864 | 753,291,867 | 864 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-09 05:38:58 +0100
From: @MarcinZukowski
To: @njnes
Version: -- development
Last updated: 2006-02-10 12:42:38 +0100
## Comment 5011
Date: 2006-02-09 17:38:58 +0100
From: @MarcinZukowski
Here's a script:
<<<
module(radix);
dir("killer_bat");
destroy("killer_bat");
commit();
dir("killer_bat");
var b :=
uniform(1024).persists(true).rename("killer_bat");
dir("killer_bat");
bat("killer_bat").count().print();
commit();
<<<
Run it once, it works.
Run it twice, it fails.
Apparently, a destroyed bat, once re-created, although
appearing in a catalog, cannot be directly referenced
by name.
## Comment 5012
Date: 2006-02-10 10:26:13 +0100
From: @njnes
Logged In: YES
user_id=43556
test added to
tests/BugTracker/Tests/bat_destroy.SF-1428446.milS and a -2
version.
## Comment 5013
Date: 2006-02-10 12:42:38 +0100
From: @njnes
Logged In: YES
user_id=43556
fixed, the backup of bat names should only keep temp names.
This wasn't done correct in the BBPinit function.
## Comment 5014
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1428446 at http://sourceforge.net/support/tracker.php?aid=1428446
| BAT destroy problem | https://api.github.com/repos/MonetDB/MonetDB/issues/863/comments | 0 | 2020-11-30T09:03:04Z | 2024-06-27T11:09:04Z | https://github.com/MonetDB/MonetDB/issues/863 | 753,291,821 | 863 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-09 05:17:25 +0100
From: Wouter Alink <<alink>>
To: @njnes
Version: -- development
Last updated: 2006-02-10 10:00:09 +0100
## Comment 5006
Date: 2006-02-09 17:17:25 +0100
From: Wouter Alink <<alink>>
When Mserver is started with:
Mserver
module(pathfinder);
pfstart();
And on the client side you try to open a MapiClient MIL
session the MapiClient hangs. On the server side:
!ERROR: xchange_challenge: frontend not found
appears. This seems to be correct (as i didn't do
mapi_start(); ), but I would expect the MapiClient not
to 'hang', but exit with a message like "MIL not
supported" or something similar.
## Comment 5007
Date: 2006-02-09 17:36:32 +0100
From: @grobian
Logged In: YES
user_id=963970
to be more exact, the message you get on the console should
go to the client instead and all would be well.
## Comment 5008
Date: 2006-02-10 08:36:52 +0100
From: @yzchang
Logged In: YES
user_id=341633
I use CVS head. MapiClient does not hang, but exits with:
MAPI = anonymous@localhost:50000
ACTION= read_line
ERROR = Connection terminated
Pretty vague:) Would be better if MapiClient exists with:
...
ERROR = frontend not found
## Comment 5009
Date: 2006-02-10 10:00:09 +0100
From: @njnes
Logged In: YES
user_id=43556
the errors are now send back to the client, ie closing this
bug. A test script is added to the multiple_server test (in
sql).
## Comment 5010
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1428431 at http://sourceforge.net/support/tracker.php?aid=1428431
| MAPI: MapiClient hangs when sending invalid challenge | https://api.github.com/repos/MonetDB/MonetDB/issues/862/comments | 0 | 2020-11-30T09:02:59Z | 2024-06-27T11:09:03Z | https://github.com/MonetDB/MonetDB/issues/862 | 753,291,755 | 862 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-09 04:17:44 +0100
From: @MarcinZukowski
To: @njnes
Version: -- development
Last updated: 2006-02-20 11:18:40 +0100
## Comment 4999
Date: 2006-02-09 16:17:44 +0100
From: @MarcinZukowski
I wanted to create a large BAT, larger than the
available memory size.
I used the approach comparable to the one performed by
bigbat() function in provided killer.mx module:
- I use BATnew with a huuge number
- I fill in the data
- I set the size with BATsetcount
- and that's it
I enforce Mserver to treat such a BAT as mem-mapped by
calling it with
--set gdk_mem_bigsize=1073741824
--set gdk_vm_minsize=1073741824
options.
I test everything on CWI cluster sn-2 machine, might
behave differently somewhere else.
The created BAT is memory mapped, as I can see in the
debugger.
Now, two problems occur:
- during BAT creation the Mserver takes hell a lot of
memory, and the operating system tends to kill it. I
don't know when it does and when not, it seems quite
random. I managed to successfully run killer_init() in
like 4-th attempt. You can repro it by doing
echo "module(killer);killer_init();" | $MSERVER
- even if the BAT creation succeeds, a simple operation
on it (on a new Mserver) results in errors. This is an
output of:
echo "module(killer);killer_test();" | $MSERVER
<<<
GDKvmalloc(17179869200) fail => BBPtrim(enter)
usage[mem=3894376,vm=118816768]
!mallinfo.arena = 1241088
!mallinfo.ordblks = 35
!mallinfo.smblks = 4
!mallinfo.hblkhd = 0
!mallinfo.hblks = 0
!mallinfo.usmblks = 0
!mallinfo.fsmblks = 192
!mallinfo.uordblks = 1065424
!mallinfo.fordblks = 175664
BBPTRIM_ENTER: memsize=3894376,vmsize=118816768
BBPTRIM: memtarget=4611686018427387904
vmtarget=4611686018427387904
TRIMSCAN: mem=1 vm=1, start=1, limit=1
TRIMSCAN: end at 1 (size=103)
TRIMSELECT: dirty = 0
TRIMSELECT: end
TRIMSELECT: dirty = 1
TRIMSELECT: end
BBPTRIM: no more unload candidates!
BBPTRIM_EXIT: memsize=4014104,vmsize=118816768
GDKvmalloc(17179869200) fail => BBPtrim(ready)
usage[mem=3894376,vm=118816768]
!mallinfo.arena = 1241088
!mallinfo.ordblks = 35
!mallinfo.smblks = 4
!mallinfo.hblkhd = 0
!mallinfo.hblks = 0
!mallinfo.usmblks = 0
!mallinfo.fsmblks = 192
!mallinfo.uordblks = 1065424
!mallinfo.fordblks = 175664
!FATAL: GDKvmalloc: failed for 16 bytes
<<<
Hence the BAT is unusable.
Perhaps I'm doing something wrong with a BAT creation,
but it seems to be fine with small BATs. I know there
is (supposedly) a thread that tries to tell OS that
some virtual memory is not used. Perhaps it reads some
information from the BAT header, which I don't update.
But I also had that problem in a scenario when I did
BATsetcount often.
I assign it to Niels as Martin suggested that.
## Comment 5000
Date: 2006-02-09 16:27:19 +0100
From: @MarcinZukowski
Logged In: YES
user_id=607094
Is probably related to the second problem.
I just experimented with killer_init with 256M and 1M values
(don't try lower, it has to be a multiple of 1M otherwise it
crashes:)).
Interestingly, calling info() on created BATs takes MUCH
longer on 256M version, meaning, that most probably, MonetDB
tries to load it from the disk, which (afaik) shouldn't happen.
## Comment 5001
Date: 2006-02-10 13:47:21 +0100
From: @MarcinZukowski
Logged In: YES
user_id=607094
It seems the problem disappears if I make the
freshly-created BAT read-only. This has to do with the
MAP_PRIVATE flag sent to mmap(), which apparently causes a
failure with large files on Fedora4.
## Comment 5002
Date: 2006-02-10 13:54:02 +0100
From: @njnes
Logged In: YES
user_id=43556
The second problem seems fixed. During the second run the
bat wasn't mmapped but malloced causing the out of memory
problems. The problem was that during HEAP creation (first
run) the bat was nicely created memory mapped. But the BATsave
overwrote the storage type because the batMapbuns wasn't set
to -1.
## Comment 5003
Date: 2006-02-20 10:13:54 +0100
From: @peterboncz
Logged In: YES
user_id=591107
right, before I forget, it seems that Fedora4 has switched
to a super-pessimistic memory mapping policy.
that is, if a file is privately mapped (which means that all
changed pages will seamlessly be replaced by pages from
swapspace), it now claims this swap-space in advance,
assuming that *all* pages are going to be changed.
This is very unlikely in MonetDB, were writable bats need to
be memory mapped privately (to protect their integrity under
crashes). Thus, large writable memory mapped bats can now
easily cause mememory failure on Fedora Core 4. Bats larger
than your swap-space will always cause such failure. This is
the first time I have ssen such a policy in an OS.
maybe there is a new mmap option in Fedora core 4 that
allows to override this new policy? did not check yet.
## Comment 5004
Date: 2006-02-20 11:18:40 +0100
From: @peterboncz
Logged In: YES
user_id=591107
I added the MAP_NORESERVE flag. That should do it.
## Comment 5005
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1428389 at http://sourceforge.net/support/tracker.php?aid=1428389
| Problem with large mem-mapped files | https://api.github.com/repos/MonetDB/MonetDB/issues/861/comments | 0 | 2020-11-30T09:02:55Z | 2024-06-27T11:09:02Z | https://github.com/MonetDB/MonetDB/issues/861 | 753,291,701 | 861 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-09 03:53:44 +0100
From: thijs <<thijswes>>
To: GDK devs <<bugs-common>>
Version: -- development
Last updated: 2006-02-09 10:52:35 +0100
## Comment 4996
Date: 2006-02-09 15:53:44 +0100
From: thijs <<thijswes>>
When seqbase is set on a [void,oid] bat, somehow the
bat is converted to [oid,oid]. It doesn't show this in
the properties or when printed, but Monet complains
when inserting a [void,oid] tuple:
ERROR: interpret: no matching MIL operator to
'insert(BAT[oid,oid], void, oid)'.
Example:
var tmp:=new(void,oid).seqbase(oid(0));
tmp.insert(nil,oid(0));
Example2:
var tmp:=new(void,oid).reverse().mark(oid(0)).reverse();
tmp.insert(nil,oid(0));
## Comment 4997
Date: 2006-02-09 22:52:35 +0100
From: @njnes
Logged In: YES
user_id=43556
This is a feature. Setting the seqbase makes logically an oid
column out of the void (nil) column. So the remaining is could
be a problem that bat info gives the wronge type (void
instead of oid). But as info basically is used for
debugging, giving
the reall low level type is more insight full.
## Comment 4998
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1428369 at http://sourceforge.net/support/tracker.php?aid=1428369
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-common@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
| insert(BAT[void,oid],void,oid) error after setting seqbase | https://api.github.com/repos/MonetDB/MonetDB/issues/860/comments | 0 | 2020-11-30T09:02:52Z | 2024-06-28T07:36:49Z | https://github.com/MonetDB/MonetDB/issues/860 | 753,291,653 | 860 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-09 01:57:41 +0100
From: e nijhuis <<nijhuis>>
To: Fabian Groffen <<fabian>>
Version: -- development
Last updated: 2006-02-09 09:14:44 +0100
## Comment 4993
Date: 2006-02-09 01:57:41 +0100
From: e nijhuis <<nijhuis>>
Hi,
I just downloaded the latest release of monetdb
(before I was running the version 4.8.0) and to my
surprise I am experiencing problems to connect to the
databse with the jdbc driver.
I am running:
Monet Database Server V4.10.0
Compiled for i686-pc-win32/32bit with 32bit OIDs;
dynamically linked.
With the foloowing java version:
java version "1.5.0_06"
Java HotSpot(TM) Client VM (build 1.5.0_06-b05, mixed
mode, sharing)
And get the following error while trying to connect
to the database:
>java -jar MonetDB_JDBC-1.2-Blunt-SR1.jar -umonetdb
password:
Exception in thread "main" java.lang.AssertionError:
Illegal start of block
at
nl.cwi.monetdb.jdbc.MonetSocketBlockMode.readLine
(Unknown Source)
at
nl.cwi.monetdb.jdbc.MonetSocket.waitForPrompt(Unknown
Source)
at nl.cwi.monetdb.jdbc.MonetConnection.<init>
(Unknown Source)
at nl.cwi.monetdb.jdbc.MonetDriver.connect
(Unknown Source)
at java.sql.DriverManager.getConnection
(Unknown Source)
at java.sql.DriverManager.getConnection
(Unknown Source)
at JdbcClient.main(Unknown Source)
Could you help me out here? Am I using an outdated
driver? (I couldn't find a newer one on your website)
Thank you for your help in advance,
Emil Nijhuis
## Comment 4994
Date: 2006-02-09 09:14:44 +0100
From: @grobian
Logged In: YES
user_id=963970
The installers and source balls include a version of the
JDBC driver that works with 4.10. Previous versions of the
driver are indeed incompatible with the latest release.
On windows it should be in "Program
Files\CWI\MonetDB\share\MonetDB\lib\monetdb-1.3-jdbc.jar"
Thanks for your report, let me know if you can't find it.
## Comment 4995
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1428011 at http://sourceforge.net/support/tracker.php?aid=1428011
| jdbc driver isn't compatible with monetdb 4.10.0 anymore? | https://api.github.com/repos/MonetDB/MonetDB/issues/859/comments | 0 | 2020-11-30T09:02:48Z | 2024-06-28T07:36:49Z | https://github.com/MonetDB/MonetDB/issues/859 | 753,291,620 | 859 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-08 10:28:42 +0100
From: @grobian
To: @sjoerdmullender
Version: -- development
Last updated: 2006-02-10 04:08:15 +0100
## Comment 4989
Date: 2006-02-08 10:28:42 +0100
From: @grobian
The monetdb configure scripts thinks somehow it should
install php-modules into
$prefix/lib/php/extensions/[debug-etc-etc]/, however,
this is not the right location:
pegatoo ~ php-config --extension-dir
/usr/lib/php5/lib/php/extensions/no-debug-non-zts-20041030
pegatoo ~ equery files monetdb | grep monetdb.so
/usr/lib/php/extensions/no-debug-non-zts-20041030/monetdb.so
/usr/lib/php/extensions/no-debug-non-zts-20041030/monetdb.so.0
/usr/lib/php/extensions/no-debug-non-zts-20041030/monetdb.so.0.0.0
The result is that a php installation that doesn't use
this default (like php5) the module is not available.
I think configure should use `php-config
--extension-dir` to determine the location of the
module. I have no idea how, hence this bug.
## Comment 4990
Date: 2006-02-09 11:03:07 +0100
From: @grobian
Logged In: YES
user_id=963970
Niels and I came to the conclusion that this is wrongly
installed files are due to the fact that the php
installation was done with another prefix than monetdb.
Gentoo has chosen in this situation to change the prefix of
php5 so it can be slotted with php4. The proper solution
would probably be to have a --with-php-extdir= directive for
configure such that I can just set it to the right location
on Gentoo ebuild installs.
## Comment 4991
Date: 2006-02-10 16:08:15 +0100
From: @njnes
Logged In: YES
user_id=43556
fixed ie added the --with-php-extensiondir option to configure.
## Comment 4992
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1427251 at http://sourceforge.net/support/tracker.php?aid=1427251
| PHP: module installed in incorrect location | https://api.github.com/repos/MonetDB/MonetDB/issues/858/comments | 0 | 2020-11-30T09:02:45Z | 2024-06-27T11:08:59Z | https://github.com/MonetDB/MonetDB/issues/858 | 753,291,571 | 858 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-07 03:59:57 +0100
From: @Teggy
To: Fabian Groffen <<fabian>>
Version: -- development
Last updated: 2006-02-07 07:28:53 +0100
## Comment 4985
Date: 2006-02-07 15:59:57 +0100
From: @Teggy
On http://monetdb.cwi.nl/XQuery/index.html, the link
to the API documentation points to
http://monetdb.cwi.nl/XQuery/Documentation/Interfaces
which yields nothing but a 404 error.
Cheers,
--Teggy
## Comment 4986
Date: 2006-02-07 19:28:26 +0100
From: @grobian
Logged In: YES
user_id=963970
supposedly fixed in CVS
## Comment 4987
Date: 2006-02-07 19:28:53 +0100
From: @grobian
Logged In: YES
user_id=963970
oops, sorry for the bugspam
## Comment 4988
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1426166 at http://sourceforge.net/support/tracker.php?aid=1426166
| Link on MonetDB/XQuery website broken | https://api.github.com/repos/MonetDB/MonetDB/issues/857/comments | 0 | 2020-11-30T09:02:42Z | 2024-06-27T11:08:58Z | https://github.com/MonetDB/MonetDB/issues/857 | 753,291,527 | 857 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-07 02:49:33 +0100
From: @sjoerdmullender
To: Jan Rittinger <<jan.rittinger>>
Version: -- development
Last updated: 2006-06-20 09:34:30 +0200
## Comment 4980
Date: 2006-02-07 14:49:33 +0100
From: @sjoerdmullender
Since the merge of the update branch into the main
branch, a number of tests fail because they trigger
code that uses KIND_PRE. KIND_PRE does not exist
anymore and has been replaced with KND_NID*. Likewise,
PROP_PRE* does not exist anymore and has been replaced
with KND_PROP*.
There is even one test where the test code contains
KIND_PRE
(tests/BugTracker/Tests/SCJ_with_void_void_BATs.SF-1400473.milS)
Assigning to Jan since it's his code that now fails and
because Stefan is still away on vacation. (Stefan was
responsible for changing the indices.)
## Comment 4981
Date: 2006-02-19 15:21:51 +0100
From: @drstmane
Logged In: YES
user_id=572415
replaced left-over occurances of old KIND_PRE_* & PROP_PRE_*
indices
by new KND_NID_* and KND_PROP_* indices.
several tests now work, again;
others still give wrong results;
further analysis is required, but only reasonably possible
once http://sourceforge.net/support/tracker.php?aid=1433507 "PF: shred_doc fails with BATproperty error"
is fixed...
## Comment 4982
Date: 2006-06-01 15:34:23 +0200
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
I think this one is outdated and should be removed.
## Comment 4983
Date: 2006-06-20 09:34:29 +0200
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
no test
## Comment 4984
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1426110 at http://sourceforge.net/support/tracker.php?aid=1426110
| XQuery: KIND_PRE* is gone | https://api.github.com/repos/MonetDB/MonetDB/issues/856/comments | 0 | 2020-11-30T09:02:39Z | 2024-06-27T11:08:57Z | https://github.com/MonetDB/MonetDB/issues/856 | 753,291,479 | 856 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-07 11:19:09 +0100
From: Wouter Alink <<alink>>
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-02-20 06:54:45 +0100
## Comment 4975
Date: 2006-02-07 11:19:09 +0100
From: Wouter Alink <<alink>>
Given the query:
let $a := <a b="10"/>
for $b in $a
return element { "d" } { $b }
returns (using 0.11):
ERROR: mposjoin: all BATs in the tail of the third
input BAT (ws_item) must have a dense head.
ERROR: CMDmposjoin: operation failed.
While the following queries return the expected result:
let $a := <a b="10" c="12"/>
for $b in $a
return element { "d" } { $b }
returns: <a c="12" b="10"/>
let $a := <a/>
for $b in $a
return element { "d" } { $b }
returns: <a/>
note: might be related to [ 1376942 ] Shredding 1
attribute
## Comment 4976
Date: 2006-02-20 11:23:17 +0100
From: @peterboncz
Logged In: YES
user_id=591107
I will try to fix this for the 0.10 bugfix release.
## Comment 4977
Date: 2006-02-20 11:33:09 +0100
From: @Teggy
Logged In: YES
user_id=642632
Note: the expected results of the two latter queries should
rather read
<d><a c="12" b="10"/></d>
and
<d><a/></d>
(note the enclosing 'd' element in both cases).
Regards,
--Teggy
## Comment 4978
Date: 2006-02-20 18:54:45 +0100
From: @peterboncz
Logged In: YES
user_id=591107
this was a bug report for the HEAD -- no issue for the
release.
anyway, in both branches, it appears that it has gone by
itself (?).
to make sure that it does not return (unnoticed), I added
it as a test script in BugsViaSourceForge
## Comment 4979
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1425978 at http://sourceforge.net/support/tracker.php?aid=1425978
| XQuery: querying 1 attribute | https://api.github.com/repos/MonetDB/MonetDB/issues/855/comments | 0 | 2020-11-30T09:02:36Z | 2024-06-28T07:36:48Z | https://github.com/MonetDB/MonetDB/issues/855 | 753,291,449 | 855 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-03 06:29:23 +0100
From: @MarcinZukowski
To: @njnes
Version: -- development
Last updated: 2006-02-13 02:01:20 +0100
## Comment 4969
Date: 2006-02-03 18:29:23 +0100
From: @MarcinZukowski
Here's a script:
<<<
PROC pro() : bat[void,int]
{
new(void,int);
printf("vm_cursize=%lld\n", vm_cursize());
}
var i := 0;
while(true) {
pro();
}
<<<
If you run it and watch vm_cursize output, you'll see
it growing (give it a second if it doesn't).
The output of the valgrind for a run with 20K iterations:
321,712 (321,152 direct, 560 indirect) bytes in 20,041
blocks are defi
nitely lost in loss record 8 of 9
==24286== at 0x4904A66: malloc (vg_replace_malloc.c:149)
==24286== by 0x5913F51: GDKmallocmax (gdk_utils.mx:1056)
==24286== by 0x59140DC: GDKmalloc (gdk_utils.mx:1077)
==24286== by 0x5914645: GDKstrdup (gdk_utils.mx:1232)
==24286== by 0x5882AA3: BBPinsert (gdk_bbp.mx:985)
==24286== by 0x596D440: BATcreatedesc (gdk_bat.mx:128)
==24286== by 0x596DBB8: BATnewstorage (gdk_bat.mx:217)
==24286== by 0x596E057: BATnew (gdk_bat.mx:266)
==24286== by 0x4E519AF: CMDnew (bat.mx:587)
==24286== by 0x4E51A27: CMDnew_default (bat.mx:598)
==24286== by 0x4E4A36F:
CMDnew_default_unpack1793450882 (bat.glue.c:74)
==24286== by 0x4A1BE7C: interpret
(monet_interpreter.mx:1172)
==24286== by 0x4A24D33: interpret_seqblock
(monet_interpreter.mx:1820)
==24286== by 0x4A19B05: interpret
(monet_interpreter.mx:766)
==24286== by 0x4A1B7FE: interpret
(monet_interpreter.mx:1155)
==24286== by 0x4A24D33: interpret_seqblock
(monet_interpreter.mx:1820)
==24286== by 0x4A19B05: interpret
(monet_interpreter.mx:766)
==24286== by 0x4A1D6A8: interpret_while
(monet_interpreter.mx:1244)
==24286== by 0x4A19DFF: interpret
(monet_interpreter.mx:797)
==24286== by 0x4A24D33: interpret_seqblock
(monet_interpreter.mx:1820)
==24286== by 0x4A19B05: interpret
(monet_interpreter.mx:766)
==24286== by 0x4A39568: handleRequest
(monet_queue.mx:536)
==24286== by 0x4A399F6: doRequest (monet_queue.mx:562)
==24286== by 0x4A8FA3F: monetInterpreter
(monet_process.mx:111)
==24286== by 0x402638: main (Mserver.mx:387)
==24286==
It seems to be leaking on the new 'bak' field,
introduced (according to CVS) by Martin.
## Comment 4970
Date: 2006-02-07 23:34:04 +0100
From: @njnes
Logged In: YES
user_id=43556
In the BBPinsert the bak field was reset to NULL before a
string could be reused.
Test added
tests/BugTracker/Tests/leak_of_backup_names.SF-1423568.milS
## Comment 4971
Date: 2006-02-10 23:09:14 +0100
From: @MarcinZukowski
Logged In: YES
user_id=607094
Just tested on the most recent CVS version, no change.
Hence reopening.
Valgrind log for 200K runs:
==28059== 25,597,336 (25,593,744 direct, 3,592 indirect)
bytes in 199,949 block
s are definitely lost in loss record 9 of 9
==28059== at 0x4904A66: malloc (vg_replace_malloc.c:149)
==28059== by 0x5919FD9: GDKmallocmax (gdk_utils.mx:1060)
==28059== by 0x591A164: GDKmalloc (gdk_utils.mx:1081)
==28059== by 0x5889409: BBPcacheit (gdk_bbp.mx:1012)
==28059== by 0x588B028: BBPrecycle (gdk_bbp.mx:1338)
==28059== by 0x5973C97: BATnewstorage (gdk_bat.mx:211)
==28059== by 0x5974182: BATnew (gdk_bat.mx:268)
==28059== by 0x4E54FAF: CMDnew (bat.mx:619)
==28059== by 0x4E55027: CMDnew_default (bat.mx:630)
==28059== by 0x4E4CC8F: CMDnew_default_unpack508342478
(bat.glue.c:74)
==28059== by 0x4A1BE7C: interpret (monet_interpreter.mx:1172)
==28059== by 0x4A24D33: interpret_seqblock
(monet_interpreter.mx:1820)
==28059== by 0x4A19B05: interpret (monet_interpreter.mx:766)
==28059== by 0x4A1B7FE: interpret (monet_interpreter.mx:1155)
==28059== by 0x4A24D33: interpret_seqblock
(monet_interpreter.mx:1820)
==28059== by 0x4A19B05: interpret (monet_interpreter.mx:766)
==28059== by 0x4A1D6A8: interpret_while
(monet_interpreter.mx:1244)
==28059== by 0x4A19DFF: interpret (monet_interpreter.mx:797)
==28059== by 0x4A24D33: interpret_seqblock
(monet_interpreter.mx:1820)
==28059== by 0x4A19B05: interpret (monet_interpreter.mx:766)
==28059== by 0x4A39568: handleRequest (monet_queue.mx:536)
==28059== by 0x4A399F6: doRequest (monet_queue.mx:562)
==28059== by 0x4A8FAAB: monetInterpreter
(monet_process.mx:111)
==28059== by 0x402638: main (Mserver.mx:391)
==28059==
## Comment 4972
Date: 2006-02-10 23:32:32 +0100
From: @MarcinZukowski
Logged In: YES
user_id=607094
Just noticed that the new valgrind run is actually different
from the one I initially submitted. If it's correct
(sometimes valgrind goes crazy), that means last fix broke
something else... Especially the difference in per-block
memory leak is interesting, it increased from ca. 16 bytes
to ca. 125 :(
## Comment 4973
Date: 2006-02-13 14:01:20 +0100
From: @MarcinZukowski
Logged In: YES
user_id=607094
Seems the last weekend changes fixed it.
## Comment 4974
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1423568 at http://sourceforge.net/support/tracker.php?aid=1423568
| Leaks in the new BAT creation | https://api.github.com/repos/MonetDB/MonetDB/issues/854/comments | 0 | 2020-11-30T09:02:33Z | 2024-06-27T11:08:55Z | https://github.com/MonetDB/MonetDB/issues/854 | 753,291,393 | 854 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-03 02:13:34 +0100
From: @MarcinZukowski
To: @njnes
Version: -- development
Last updated: 2006-02-10 02:59:18 +0100
## Comment 4964
Date: 2006-02-03 14:13:34 +0100
From: @MarcinZukowski
Here's the problem:
<<
module("radix");
works
var b:=uniform(1000).access(BAT_WRITE);
b.append(b);
b.count().print();
works
b:=uniform(1000).reverse().mark().reverse().access(BAT_APPEND);
b:=b.append(b.copy());
b.count().print();
doesn't
b:=uniform(1000).reverse().mark().reverse().access(BAT_APPEND);
b:=b.append(b);
b.count().print();
<<<
To me it looks like a bug.
## Comment 4965
Date: 2006-02-10 14:59:18 +0100
From: @njnes
Logged In: YES
user_id=43556
Its more a feature. Accessing the same bat concurently is
protected within the mil interpreter. For the second bat a
view is created. Then the first cannot be changed anymore.
So why a feature. If you concurrently append and read from a
bat you are basically creating an infinite bat as new values
have been added which again are added etc etc.
So I'll close this with won't fix. And no test is added too.
## Comment 4966
Date: 2006-02-10 16:36:24 +0100
From: @MarcinZukowski
Logged In: YES
user_id=607094
Well, I have to disagree.
Appending a bat to itself clearly works, look at the first
example in the bug sumbmission. So the argument about
infinite bat is invalid.
It IS possible to append a BAT for which a view is created:
var x := new(void,int).append(1);
var y := x.reverse().mark(0@0).reverse();
x.append(2); works
The problem is, the view on a bat cannot be modified.
y.append(3); doesn't work
So it boils down to the fact, that you cannot append a BAT
to itself if there is a view on it.
x.append(x); doesn't work
But
x.append(y); works
And that I find incorrect. If you can append a bat to
itself, and append a value/other bat to a bat with a view on
it, then why not the bat with a view on it to itself.
Unless I don't get something. Again :)
## Comment 4967
Date: 2006-02-10 16:47:26 +0100
From: @njnes
Logged In: YES
user_id=43556
b.append(b) of case one isn't appending it self to b, but a
copy. This is because of the same pining, as b is writable
views aren't allowed, ie the second b is no view bat a copy.
Same holds for you last new example (append(2)), Y is a copy
not a view on x.
So it boils down to the fact, that you cannot append a BAT
to itself if it is append able. I aggree this is strange,
but it can only be solved by removing the pining stuff which
is needed for concurrent void (virtual ids have no pointer,
so we use a single field in the bat) access.
## Comment 4968
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1423366 at http://sourceforge.net/support/tracker.php?aid=1423366
| Append and bat access problem | https://api.github.com/repos/MonetDB/MonetDB/issues/853/comments | 0 | 2020-11-30T09:02:30Z | 2024-06-28T07:36:48Z | https://github.com/MonetDB/MonetDB/issues/853 | 753,291,352 | 853 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-02-03 01:52:35 +0100
From: @yzchang
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-02-22 10:36:35 +0100
## Comment 4958
Date: 2006-02-03 13:52:35 +0100
From: @yzchang
Run the following query:
====
declare namespace local = "http://www.foobar.org";
declare function local:convert($v as xs:double?) as
xs:double?
{
2.20371 * $v
};
local:convert(25)
====
and it will result in an error:
$mapiclnt -lx q18.xq
ERROR: interpret: unknown variable 'item001'.
ERROR: interpret_params: leftfetchjoin(param 2):
evaluation error.
But all of the following queries do _not_ fail (note
the changes in the
function declaration):
1. change the OccurrenceIndicator
====
declare namespace local = "http://www.foobar.org";
declare function local:convert($v as xs:double) as
xs:double?
{
2.20371 * $v
};
local:convert(25)
====
====
declare namespace local = "http://www.foobar.org";
declare function local:convert($v as xs:double?) as
xs:double
{
2.20371 * $v cast as xs:double
};
local:convert(25)
====
2. change the parameter/return_value type
====
declare namespace local = "http://www.foobar.org";
declare function local:convert($v as xs:decimal?) as
xs:double?
{
2.20371 * $v cast as xs:double
};
local:convert(25)
====
====
declare namespace local = "http://www.foobar.org";
declare function local:convert($v as xs:double?) as
xs:decimal?
{
2.20371 * $v cast as xs:decimal
};
local:convert(25)
====
3. the query won't fail, if we add a useless cast:
====
declare namespace local = "http://www.foobar.org";
declare function local:convert($v as xs:double?) as
xs:double?
{
2.20371 * $v cast as xs:double
};
local:convert(25)
====
After having played around, I noticed that only when
the first query
(the one causes failure) is translated into mil, the
function
"evaluateOpOpt" in milprint_summer.c is used, where
".leftfetchjoin(item001)" is generated.
During the translation to mil process of all other
queries, the function
"evaluateOp" is used, where no mil code like
".leftfetchjoin(item001)"
is generated.
## Comment 4959
Date: 2006-02-20 11:22:34 +0100
From: @peterboncz
Logged In: YES
user_id=591107
I will try to fix this for the 0.10 bugfix release.
## Comment 4960
Date: 2006-02-20 18:57:41 +0100
From: @peterboncz
Logged In: YES
user_id=591107
fix of character edit distance 1 (bitwise hamming-distance
2) in milprint_summer.c in the routine indicated by Jennie.
## Comment 4961
Date: 2006-02-21 12:59:16 +0100
From: @drstmane
Logged In: YES
user_id=572415
re-opened to remind us that we need to add a test script to CVS.
## Comment 4962
Date: 2006-02-22 10:36:35 +0100
From: @drstmane
Logged In: YES
user_id=572415
Peter added test
pathfinder/tests/BugsViaSourgeforce/Tests/ID.1423349.xq
## Comment 4963
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1423349 at http://sourceforge.net/support/tracker.php?aid=1423349
| PF: err when "xs:double" is used | https://api.github.com/repos/MonetDB/MonetDB/issues/852/comments | 0 | 2020-11-30T09:02:26Z | 2024-06-27T11:08:53Z | https://github.com/MonetDB/MonetDB/issues/852 | 753,291,306 | 852 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-30 08:45:53 +0100
From: mihalych <<mihalych_>>
To: Fabian Groffen <<fabian>>
Version: -- development
Last updated: 2006-01-30 10:42:35 +0100
## Comment 4955
Date: 2006-01-30 20:45:53 +0100
From: mihalych <<mihalych_>>
Connection is switching to autocommit mode after
commit() or rollback(), but it should not.
## Comment 4956
Date: 2006-01-30 22:42:35 +0100
From: @grobian
Logged In: YES
user_id=963970
I have fixed this in the stable branch. Please grab the
nightly stable build sources tomorrow and see if it works
like you expect.
Added test in sql/src/jdbc/tests/Test_Ctransaction.java
## Comment 4957
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1418966 at http://sourceforge.net/support/tracker.php?aid=1418966
| JDBC connection's autocommit state | https://api.github.com/repos/MonetDB/MonetDB/issues/851/comments | 0 | 2020-11-30T09:02:23Z | 2024-06-27T11:08:52Z | https://github.com/MonetDB/MonetDB/issues/851 | 753,291,268 | 851 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-25 10:21:42 +0100
From: Wouter Alink <<alink>>
To: @sjoerdmullender
Version: -- development
Last updated: 2006-02-24 03:31:43 +0100
## Comment 4946
Date: 2006-01-25 22:21:42 +0100
From: Wouter Alink <<alink>>
Trying the 0.10 XQuery build...
First starting the XQuery server:
Monet Database Server V4.10.0
Copyright (c) 1993-2006, CWI. All rights reserved.
Compiled for i386-pc-mingw32/32bit with 32bit OIDs;
dynamically linked.
Visit http://monetdb.cwi.nl/ for further
information.
MonetDB>!WARNING: MAPIlisten: Unix domain sockets are
not supported
Although there is the warning, starting a MapiClient
MIL session does work. But starting a MapiClient
XQuery session (from the Start Menu) does not work.
(a screen pops up and directly closes)
Why not... the answer is quite simple: XQuery is not
interactive (at least not when using MapiClient). I
suggest either starting the JDBCClient (which is
interactive) or remove the XQueryClient menu-item.
## Comment 4947
Date: 2006-01-25 22:29:15 +0100
From: Wouter Alink <<alink>>
Logged In: YES
user_id=621590
sorry, was to soon to draw conclusions. of course the
start-menu-item should start a mil-session and not an
xquery session, but it does not seem to do so. it looks
like the lnk-file does not point to MapiClient.bat?? i'm
confused now...
## Comment 4948
Date: 2006-01-25 22:35:12 +0100
From: Wouter Alink <<alink>>
Logged In: YES
user_id=621590
found the problem: mapi_start(); is not called in the
initialization script of MPFserver.
## Comment 4949
Date: 2006-01-27 10:46:43 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
The initial report about "!WARNING: MAPIlisten: Unix domain
sockets are not supported" is valid. I didn't have time to
investigate and fix that before the release.
But why do you think that starting an XQuery server should
start a MIL server? It shouldn't. It never has. If you
think otherwise, submit a feature request, but don't start
adding features to the stable branch.
I am reverting the checkin and reopening the bug, since the
original problem is then still there.
## Comment 4950
Date: 2006-01-27 10:51:13 +0100
From: @grobian
Logged In: YES
user_id=963970
I think Wouter wants to be able to do shred_doc or
something. Related to that I'd like to be able to do such
thing via the XQuery interface by piping a document into it,
instead of needing local filesystem access. I guess this is
a feature request... ok useless comment here.
## Comment 4951
Date: 2006-01-27 16:19:18 +0100
From: Wouter Alink <<alink>>
Logged In: YES
user_id=621590
I don't think I added a new feature (there was already a
menu-item that wanted to open a MIL-session, only it did not
do anything, so that's why i tried to solve a bug). It was
not my intention at least to add a new feature.
A MIL-client is (to my best knowledge) currently still
needed to shred documents to the 'persistent document
collection', and therefore my opinion is that a MIL-session
should be part of the XQuery-product, until there is a
better solution to it. I agree with fabian that there should
be another way to add documents.
Note: although there is a warning when typing
"mapi_start();", starting a MIL-client afterwards seems to
work fine. (at least on my windows pc)
## Comment 4952
Date: 2006-01-27 16:53:25 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Try not to confuse things and make separate reports for
separate issues. Don't give your *interpretation* of what
goes wrong, just *tell* what goes wrong. Read
http://www.chiark.greenend.org.uk/~sgtatham/bugs.html and
read it again.
This report started out with
!WARNING: MAPIlisten: Unix domain sockets are not supported
That's a valid report.
However, the rest of your report is actually erroneous: no
MIL server is started, so no MIL session in a new MapiClient
can be started. (You do get the Mserver console where you
can type MIL commands.) The menu entry you refer to
attempts to do just that (and fails). The fact that you
can't have interactive XQuery sessions has nothing to do
with this.
This warning is not the same thing as "there is a MonetDB
Client entry in the Start menu, but it doesn't do anything."
That warrants another bug report. It is your interpretation
that these two have anything to do with each other (they
don't: I get the same warning when I call mapi_start()--your
attempted fix.).
## Comment 4953
Date: 2006-02-24 15:31:42 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Closed since the only actual error has been fixed: there is
no longer a complaint that Unix domain sockets aren't
supported (even though that is a correct statement :-).
## Comment 4954
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1414955 at http://sourceforge.net/support/tracker.php?aid=1414955
| XQuery: interactive Windows | https://api.github.com/repos/MonetDB/MonetDB/issues/850/comments | 0 | 2020-11-30T09:02:20Z | 2024-06-27T11:08:51Z | https://github.com/MonetDB/MonetDB/issues/850 | 753,291,225 | 850 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-25 05:14:40 +0100
From: @peterboncz
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2009-02-17 01:16:20 +0100
## Comment 4938
Date: 2006-01-25 17:14:40 +0100
From: @peterboncz
the document cache is broken.
test:
add a rule to cach http pages:
xmlcache_add_rule("http", 1000000);
then run a query
count(doc("http://www.cwi.nl/~boncz/xmark/auctions.xml")//*)
and then check in Mserver
xmlcache_print();
it will not be cached while it should have been.
## Comment 4939
Date: 2006-11-06 00:31:00 +0100
From: @drstmane
Logged In: YES
user_id=572415
Status set to "pending" as we should check, whether we can
add tests for the XML document caching features.
## Comment 4940
Date: 2007-11-06 04:20:02 +0100
From: Pseudo user for Sourceforge import <<sfimport>>
---- Original comment by: sf-robot@users.sourceforge.net ----
Logged In: YES
user_id=1312539
Originator: NO
This Tracker item was closed automatically by the system. It was
previously set to a Pending status, and the original submitter
did not respond within 365 days (the time period specified by
the administrator of this Tracker).
## Comment 4941
Date: 2007-11-06 07:06:32 +0100
From: @drstmane
Logged In: YES
user_id=572415
Originator: NO
Overruling the sf-robot:
Set status back to pending as (even after a year) we still haven't (tried to) add a respective test.
## Comment 4942
Date: 2008-11-06 03:20:11 +0100
From: Pseudo user for Sourceforge import <<sfimport>>
---- Original comment by: sf-robot@users.sourceforge.net ----
This Tracker item was closed automatically by the system. It was
previously set to a Pending status, and the original submitter
did not respond within 365 days (the time period specified by
the administrator of this Tracker).
## Comment 4943
Date: 2009-02-16 14:10:20 +0100
From: @drstmane
re-opened to remind us that we should consider adding a test
## Comment 4944
Date: 2009-02-17 01:16:20 +0100
From: @drstmane
test added in pathfinder/tests/BugTracker/Tests/XML_document_cache_broken.SF-1414720.*
appears to work fine.
## Comment 4945
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1414720 at http://sourceforge.net/support/tracker.php?aid=1414720
| XML document cache broken | https://api.github.com/repos/MonetDB/MonetDB/issues/849/comments | 0 | 2020-11-30T09:02:17Z | 2024-06-27T11:08:50Z | https://github.com/MonetDB/MonetDB/issues/849 | 753,291,175 | 849 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-25 02:36:52 +0100
From: @MarcinZukowski
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-02-19 04:45:59 +0100
## Comment 4930
Date: 2006-01-25 02:36:52 +0100
From: @MarcinZukowski
During some experiments I observed steadily increasing
MonetDB vm_cursize() allocation. I managed to trim down
the problem to the attached script:
<<<
module(radix);
var x:=uniform(10000000); to fill the memory
PROC simple(str s) : void
{
s := s + s ;
}
PROC complicated(str s) : void
{
s := s + s + s + s + s + s + s + s
+ s + s + s + s + s + s + s + s
+ s + s + s + s + s + s + s + s
+ s + s + s + s + s + s + s + s;
}
var s := "shaolin kung-fu is the best";
var vmold;
var vm := vm_cursize();
simple(s);
vmold := vm ; vm := vm_cursize();
printf("inc: %lld\n", vm - vmold);
simple(s);
vmold := vm ; vm := vm_cursize();
printf("inc: %lld\n", vm - vmold);
complicated(s);
vmold := vm ; vm := vm_cursize();
printf("inc: %lld\n", vm - vmold);
complicated(s);
vmold := vm ; vm := vm_cursize();
printf("inc: %lld\n", vm - vmold);
quit();
<<<
As a result of running
Mserver bug.mil
I get this:
inc: 1248
inc: 1596
inc: 12780
inc: 12780
Running
Mserver < bug.mil
doesn't result in memory increase.
I have some problems with reproducing it on my old
MonetDB server.
If this problem is reproducable, I'm wondering why the
memory consumption of Monet increases depending on what
function has been called. Doesn't make sense to me.
Can anyone
- try to reproduce these results?
- explain what is happening?
## Comment 4931
Date: 2006-01-25 12:48:57 +0100
From: @mlkersten
Logged In: YES
user_id=490798
The offline comments seem to indicate YYTREE management
issues.
## Comment 4932
Date: 2006-01-25 21:21:24 +0100
From: @peterboncz
Logged In: YES
user_id=591107
sure:
- reproduce = rerun, should be perfectly reproducable
- explain => function application is implemented by pasting
the syntax tree of the function body in the MIL query.
furthermore, there is a differense between
- Mserver script.mil (block mode: parses entire script into
one statement)
- Mserver < script.mil (line mode; many statements; thus
garbage collection after each statement.
so the big PRIC gets instantiated and inside the program
when the VM size is printed has not yet been deallocated (in
block mode); whereas in line mode it has been deallocated
as to why MIL is what it is; another question. But the
pasting technique can be very effective when the proc is in
a loop; expescially since the tree structure contains
interpretation and garbage collection hints that can be reused
## Comment 4933
Date: 2006-01-25 21:28:15 +0100
From: @MarcinZukowski
Logged In: YES
user_id=607094
Then another question.
Is it possible to easily enforce garbage collection for a
block of code? I tried simply putting statements in {}, but
it didn't help.
## Comment 4934
Date: 2006-02-10 16:13:30 +0100
From: @njnes
Logged In: YES
user_id=43556
Do you have an example mil script which shows your last
problem.
## Comment 4935
Date: 2006-02-10 18:33:53 +0100
From: @MarcinZukowski
Logged In: YES
user_id=607094
I don't know what you mean with the "last problem".
The script in the submission shows the problem, and I guess
Peter's comment explains what is happening. Not that I like
it, but it seems to be 'a feature' :)
## Comment 4936
Date: 2006-02-19 16:45:59 +0100
From: @peterboncz
Logged In: YES
user_id=591107
Let me clarify: MIL performs garbage collection of
intermediate results on the statement level. So, after each
statement has finished. This independent of the parsing mode
(block or line).
In block mode, an entire block of MIL (multiple statements)
is read, and a yytree is built for it. If it consists of a
lot of proc invocations, these gets pasted in and its get
even bigger.
Putting in blocks won't change the parsing strategy, so no,
that is not an option. The best performance you get when you
can rewrite your script to use a loop or a multiplex (of a
proc).
closing this issue now.
Peter
## Comment 4937
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1414208 at http://sourceforge.net/support/tracker.php?aid=1414208
| strange memory consumption | https://api.github.com/repos/MonetDB/MonetDB/issues/848/comments | 0 | 2020-11-30T09:02:14Z | 2024-06-28T07:36:48Z | https://github.com/MonetDB/MonetDB/issues/848 | 753,291,128 | 848 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-24 08:01:59 +0100
From: @grobian
To: Jens Teubner <<teubner>>
Version: -- development
Last updated: 2006-01-25 09:47:41 +0100
## Comment 4925
Date: 2006-01-24 20:01:59 +0100
From: @grobian
gcc -m32 -pipe -std=c99 -Wall -W -g
-Werror-implicit-function-declaration -Werror
-Wno-unused-function -Wno-unused-label
-Wno-uninitialized -o pf main.o ./.libs/libcompiler1.a
schema/.libs/libschema.a debug/.libs/libdebug.a
mem/.libs/libmem.a parser/.libs/libparser.a
semantics/.libs/libsemantics.a core/.libs/libcore.a
algebra/.libs/libalgebra.a mil/.libs/libmil.a
./.libs/libcompiler2.a -L/usr/lib
/usr/lib/libxml2.dylib -lz -lpthread
/usr/lib/libiconv.dylib -lm -ldl
/usr/bin/ld: multiple definitions of symbol _var_env
core/.libs/libcore.a(libcore_la-simplify.o) definition
of _var_env in section (__DATA,__common)
core/.libs/libcore.a(libcore_la-coreopt.o) definition
of _var_env in section (__DATA,__common)
collect2: ld returned 1 exit status
make[4]: *** [pf] Error 1
make[3]: *** [all-recursive] Error 1
make[2]: *** [all] Error 2
make[1]: *** [all-recursive] Error 1
make: *** [all] Error 2
I'm going to see if I can patch this. Looks like the
bison stuff works, so it would be great if we would
have an out-of-the-box install on OSX too.
## Comment 4926
Date: 2006-01-24 20:20:05 +0100
From: @grobian
Logged In: YES
user_id=963970
changing
PFarray_t *var_env[HASH_BUCKETS];
into
static PFarray_t *var_env[HASH_BUCKETS];
seems to make at least the linker happy...
## Comment 4927
Date: 2006-01-24 20:55:43 +0100
From: @grobian
Logged In: YES
user_id=963970
Jens (assuming you are the guy to ask this :) ),
I can check in my 'fix' in the burg files, but I rather do
not. I use a patch in the Gentoo ebuild distribution to
accomplish the effect of having a compiling version on OSX,
but prefer you to have a look at it and confirm that this is
indeed a proper solution.
Thanks.
## Comment 4928
Date: 2006-01-25 09:47:41 +0100
From: Jens Teubner <<teubner>>
Logged In: YES
user_id=731390
Yes, these should have been declared static.
I have just committed the changes. (I'm just
realizing that I checked in to CVS HEAD.
I will also check in the fix to the stable
branch, sorry.)
Thanks, Fabian, for tracking this down.
## Comment 4929
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1413956 at http://sourceforge.net/support/tracker.php?aid=1413956
| PF: multiple definitions of symbol _var_env on Darwin8 | https://api.github.com/repos/MonetDB/MonetDB/issues/847/comments | 0 | 2020-11-30T09:02:11Z | 2024-06-27T11:08:48Z | https://github.com/MonetDB/MonetDB/issues/847 | 753,291,091 | 847 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-24 07:05:38 +0100
From: Stratos Idreos <<idreos>>
To: Stefan Manegold <<Stefan.Manegold>>
Version: -- development
Last updated: 2006-01-25 12:18:02 +0100
## Comment 4922
Date: 2006-01-24 19:05:38 +0100
From: Stratos Idreos <<idreos>>
the following link does not work
monetdb/TechDocs/FrontEnds/XML/saxophone/index.html
## Comment 4923
Date: 2006-01-25 12:18:02 +0100
From: @drstmane
Logged In: YES
user_id=572415
fixed.
## Comment 4924
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1413906 at http://sourceforge.net/support/tracker.php?aid=1413906
| website saxophone link does not work | https://api.github.com/repos/MonetDB/MonetDB/issues/846/comments | 0 | 2020-11-30T09:02:08Z | 2024-06-27T11:08:47Z | https://github.com/MonetDB/MonetDB/issues/846 | 753,291,049 | 846 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-24 06:46:14 +0100
From: Stratos Idreos <<idreos>>
To: GDK devs <<bugs-common>>
Version: -- development
Last updated: 2006-06-06 09:23:06 +0200
## Comment 4918
Date: 2006-01-24 18:46:14 +0100
From: Stratos Idreos <<idreos>>
The xquery research roadmap page does not exist
## Comment 4919
Date: 2006-03-21 14:25:52 +0100
From: @grobian
Logged In: YES
user_id=963970
It does exist, but looks a bit minimal:
Scientific Roadmap
todo - UpdateX - Algebra - Query Optimalization
Bump!
## Comment 4920
Date: 2006-06-06 21:23:04 +0200
From: @grobian
Logged In: YES
user_id=963970
page/link seems to be removed from the website. Closing.
## Comment 4921
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1413892 at http://sourceforge.net/support/tracker.php?aid=1413892
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-common@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
| website xquery research roadmap page | https://api.github.com/repos/MonetDB/MonetDB/issues/845/comments | 0 | 2020-11-30T09:02:05Z | 2024-06-27T11:08:46Z | https://github.com/MonetDB/MonetDB/issues/845 | 753,291,009 | 845 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-24 05:29:15 +0100
From: Stratos Idreos <<idreos>>
To: GDK devs <<bugs-common>>
Version: -- development
Last updated: 2006-01-25 08:58:39 +0100
## Comment 4914
Date: 2006-01-24 17:29:15 +0100
From: Stratos Idreos <<idreos>>
In monetdb website:
monetdb/XQuery/Interface/POSIX/index.html
last link "examples" results in the following error
message:
an error occurred while processing this directive
## Comment 4915
Date: 2006-01-24 22:50:25 +0100
From: Wouter Alink <<alink>>
Logged In: YES
user_id=621590
fixed in CVS (had to point to '/XQuery/Demo/index.html')
can somebody close it?
## Comment 4916
Date: 2006-01-25 08:58:39 +0100
From: @grobian
Logged In: YES
user_id=963970
examples seem to work, closing.
## Comment 4917
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1413820 at http://sourceforge.net/support/tracker.php?aid=1413820
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-common@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
| website broken link | https://api.github.com/repos/MonetDB/MonetDB/issues/844/comments | 0 | 2020-11-30T09:02:02Z | 2024-06-27T11:08:45Z | https://github.com/MonetDB/MonetDB/issues/844 | 753,290,975 | 844 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-24 04:58:33 +0100
From: Steffen Goeldner <<sgoeldner>>
To: @njnes
Version: -- development
Last updated: 2006-02-10 04:18:38 +0100
## Comment 4906
Date: 2006-01-24 16:58:33 +0100
From: Steffen Goeldner <<sgoeldner>>
Sorry, this bug report isn't very specific.
After running the DBD::monetdb tests ca. 3 1/2 times,
the client hangs.
A netcat conversation shows:
netcat localhost 45123
16::mserver_sql:4
monetdb:monetdb
Note the missing \001\001.
The server is no longer responsive.
Need to exit();/restart the server.
## Comment 4907
Date: 2006-01-24 20:20:25 +0100
From: @njnes
Logged In: YES
user_id=43556
the DBD::monetdb tests probably make many connections per
run. So how many connections/clients do you expect this is?
## Comment 4908
Date: 2006-01-25 09:03:34 +0100
From: Steffen Goeldner <<sgoeldner>>
Logged In: YES
user_id=1174087
21 per run, but never concurrently. Many connects/
disconnects (> 100 or so) is not the problem. I.e., the
tests do something that's difficult to reproduce in a
simpler test case. Also, I cannot observe any memory/
handle leaks.
## Comment 4909
Date: 2006-01-25 11:56:20 +0100
From: Steffen Goeldner <<sgoeldner>>
Logged In: YES
user_id=1174087
Test case:
- start server
- exit();- start server
- run 136x 'c -l sql < t.sql && c < t.mil'
where c is the MapiClient and t.* are the scripts below:
t.sql:
drop table t;
create table t( c int );
t.mil:
abort();
BTW: w/o the exit(); the server crashes
## Comment 4910
Date: 2006-01-25 12:53:40 +0100
From: @njnes
Logged In: YES
user_id=43556
Doing an abort really breaks everything. Abort makes the sql
state inconsistent, ie anything can happen. I agree a hanging
client is strange. If I replace the abort with print(1);
both script run > 136 times without problems. Also I don't
get a crash if I try without the exit step inbetween.
## Comment 4911
Date: 2006-01-27 10:23:30 +0100
From: Steffen Goeldner <<sgoeldner>>
Logged In: YES
user_id=1174087
Is this abort() behavior by design?
Should I drop it from the MIL test script?
Are there alternatives?
## Comment 4912
Date: 2006-02-10 16:18:38 +0100
From: @njnes
Logged In: YES
user_id=43556
Yes the global abort behavior is by design. Also by design
mixing low level mil with sql isn't supported, and will not be
solved soon. Ie. the solution is not to mix languages (or only
very carrefully).
## Comment 4913
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1413792 at http://sourceforge.net/support/tracker.php?aid=1413792
| No server respose | https://api.github.com/repos/MonetDB/MonetDB/issues/843/comments | 0 | 2020-11-30T09:01:59Z | 2024-06-28T07:36:47Z | https://github.com/MonetDB/MonetDB/issues/843 | 753,290,928 | 843 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-22 10:56:27 +0100
From: @grobian
To: @sjoerdmullender
Version: -- development
Last updated: 2006-01-24 08:08:55 +0100
## Comment 4903
Date: 2006-01-22 22:56:27 +0100
From: @grobian
On OSX:
tar: sql-2.10.0/src/test/utf8/prov\355ncial.fmt: Cannot
open: Invalid argument
tar: sql-2.10.0/src/test/utf8/prov\355ncial.dat: Cannot
open: Invalid argument
An interesting thread on the issue:
http://www.mail-archive.com/bug-tar@gnu.org/msg00462.html
and this:
http://www.mail-archive.com/bug-tar@gnu.org/msg00487.html
The respective flag doesn't work by the way.
## Comment 4904
Date: 2006-01-24 08:08:55 +0100
From: @njnes
Logged In: YES
user_id=43556
the latin1 filenames are changed to utf8 (also some
incorrect utf8 symbols inside the files are fixed)
## Comment 4905
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1412324 at http://sourceforge.net/support/tracker.php?aid=1412324
| SQL: sourcefiles contain filenames with invalid characters | https://api.github.com/repos/MonetDB/MonetDB/issues/842/comments | 0 | 2020-11-30T09:01:56Z | 2024-06-27T11:08:43Z | https://github.com/MonetDB/MonetDB/issues/842 | 753,290,894 | 842 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-22 06:43:52 +0100
From: @yzchang
To: Stefan Manegold <<Stefan.Manegold>>
Version: -- development
Last updated: 2006-01-22 08:28:26 +0100
## Comment 4899
Date: 2006-01-22 18:43:52 +0100
From: @yzchang
Since Stefan has propagated the changes from the main
branch to the pathfinder-soap branch, I was trying to
rebuild the whole thing. But after having rebuild and
installed buildtools and MonetDB, it failed to compile
pathfinder:
../../runtime/pathfinder.mx: In function
'xquery_function_call':
../../runtime/pathfinder.mx:1377: error: 'nil'
undeclared (first use in this function)
../../runtime/pathfinder.mx:1377: error: (Each
undeclared identifier is reported only once
../../runtime/pathfinder.mx:1377: error: for each
function it appears in.)
../../runtime/pathfinder.mx:1378: error: invalid
operands to binary *
../../runtime/pathfinder.mx:1378: error: syntax error
before 'size_t'
../../runtime/pathfinder.mx:1378: error: invalid
operands to binary *
../../runtime/pathfinder.mx:1378: error: syntax error
before 'size_t'
../../runtime/pathfinder.mx:1378: error: syntax error
before 'str'
../../runtime/pathfinder.mx:1378: error: syntax error
before 'size_t'
../../runtime/pathfinder.mx: At top level:
../../runtime/pathfinder.mx:1606: error: invalid
storage class for function 'xquery_module_free'
../../runtime/pathfinder.mx:1673: error: invalid
storage class for function 'xquery_module_load'
../../runtime/pathfinder.mx:1757: error: invalid
storage class for function 'xquery_client_alloc'
../../runtime/pathfinder.mx:1799: error: invalid
storage class for function 'xquery_client_init'
../../runtime/pathfinder.mx:1878: error: invalid
storage class for function 'xquery_client_free'
../../runtime/pathfinder.mx:1936: error: invalid
storage class for function 'xquery_client_end'
../../runtime/pathfinder.mx:1992: error: invalid
storage class for function 'xquery_client_flushall'
../../runtime/pathfinder.mx:2058: error: invalid
storage class for function 'xquery_parse_ident'
../../runtime/pathfinder.mx:2072: error: invalid
storage class for function 'xquery_parse_comment'
../../runtime/pathfinder.mx:2087: error: invalid
storage class for function 'xquery_parse_space'
../../runtime/pathfinder.mx:2103: error: invalid
storage class for function 'xquery_parse_string'
../../runtime/pathfinder.mx:2173: error: invalid
storage class for function 'xquery_parse_numeric'
../../runtime/pathfinder.mx:2203: error: invalid
storage class for function 'xquery_prepare'
../../runtime/pathfinder.mx:2385: error: invalid
storage class for function 'xquery_server_end'
make[2]: *** [lib_pathfinder_la-pathfinder.lo] Error 1
make[1]: *** [all-recursive] Error 1
make: *** [all] Error 2
Can someone give me a hint how I can correct these
errors to get pathfinder compiled again.
Thanks in advance!
Jennie
## Comment 4900
Date: 2006-01-22 20:28:26 +0100
From: @drstmane
Logged In: YES
user_id=572415
Fixed compilation in CVS.
... haven't tested, though, how it runs ...
please let us know, if you have problems.
## Comment 4901
Date: 2006-01-22 20:49:11 +0100
From: @yzchang
Logged In: YES
user_id=341633
Stefan, with your fix I can compile the pathfinder soap
branch again. Thank you very much!
Jennie
## Comment 4902
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1412190 at http://sourceforge.net/support/tracker.php?aid=1412190
| pf (soap branch): failed to compile | https://api.github.com/repos/MonetDB/MonetDB/issues/841/comments | 0 | 2020-11-30T09:01:53Z | 2024-06-27T11:08:42Z | https://github.com/MonetDB/MonetDB/issues/841 | 753,290,842 | 841 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-22 09:36:14 +0100
From: @mlkersten
To: @njnes
Version: -- development
Last updated: 2007-11-28 08:13:54 +0100
## Comment 4889
Date: 2006-01-22 09:36:14 +0100
From: @mlkersten
A simple script has been added to the monet4/ backend.
It supports simple testing of concurrent access to the
server.
The call
script.sh 2 0 20 example.sql
is sufficient to crash the server.
## Comment 4890
Date: 2006-01-24 08:11:15 +0100
From: @njnes
Logged In: YES
user_id=43556
concurrent access to the cache of catalog information wasn't
save. A second problem was that the query cache wasn't
cleared after each schema update, this is now fix by
checking a global
schema number at the start of each transaction.
## Comment 4891
Date: 2006-11-06 00:25:53 +0100
From: @drstmane
Logged In: YES
user_id=572415
Status set to "pending", as we should check whether we could
add this script to the test collection.
## Comment 4892
Date: 2007-06-21 12:47:28 +0200
From: @romulogoncalves
Logged In: YES
user_id=1498628
Originator: NO
The bug needs to be open.
In M5 and M4-SQL stable crash with the script call.
I would like to move this script to the BugTracker in SQL because is not specific bug of M4,
am I allowed to do that?
Regards,
Romulo
## Comment 4893
Date: 2007-06-21 12:58:15 +0200
From: @drstmane
Logged In: YES
user_id=572415
Originator: NO
Romulo,
no need to be "shy", just go ahead, re-open the bug and make it a SQL bug, if you think that's what should be done!
Stefan
## Comment 4894
Date: 2007-06-21 13:29:16 +0200
From: @romulogoncalves
Logged In: YES
user_id=1498628
Originator: NO
The bug is open (cool now I have permissions for that :) ) and the test was added to:
/sql/src/test/BugTracker/Tests/crash_on_concurrent_use.script.SF-1411926
Regards,
Romulo
## Comment 4895
Date: 2007-11-16 13:16:57 +0100
From: @mlkersten
Logged In: YES
user_id=490798
Originator: YES
Niels, is this bug solved after your recent changes?
## Comment 4896
Date: 2007-11-28 13:23:49 +0100
From: @romulogoncalves
Logged In: YES
user_id=1498628
Originator: NO
I just ran the script (5 times) against my Mserver5 from the current cvs head branch and the mserver did not crash.
I will move the test to BugTracker and approve it and of course close the bug. However, I would like to report here what was the solution, but it seems who fixed it
forgot the normal procedures (add test in the correct place, approve the test, report that in the bug tracker and close the bug).
Any idea idea Niels?
Regards,
Romulo
## Comment 4897
Date: 2007-11-28 20:13:54 +0100
From: @romulogoncalves
Logged In: YES
user_id=1498628
Originator: NO
The bug was fixed by the Niels several checkins, but I only have track of the one done in:
04/11/2007
Log message:
fixed concurrency bug. (or simply enable the locks when needed)
The test to cover this bug is
sql/src/test/concurrent/Tests/crash_on_concurrent_use.SF-1411926
Regards,
Romulo
## Comment 4898
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1411926 at http://sourceforge.net/support/tracker.php?aid=1411926
| SQL: crash on concurrent use | https://api.github.com/repos/MonetDB/MonetDB/issues/840/comments | 0 | 2020-11-30T09:01:50Z | 2024-06-27T11:08:41Z | https://github.com/MonetDB/MonetDB/issues/840 | 753,290,794 | 840 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-22 08:53:51 +0100
From: @mlkersten
To: @sjoerdmullender
Version: -- development
Last updated: 2007-01-11 04:47:43 +0100
## Comment 4883
Date: 2006-01-22 08:53:51 +0100
From: @mlkersten
Gold demo error.
If you set TSTTRGDIR="" and run
Mserver load.mil
you get a segfault
0 0x00000037b64734b8 in memset () from /lib64/libc.so.6
1 0x00002aaaab870fa2 in GDKfree (blk=Variable "blk"
is not available.
)
at /ufs/mk/opensource/MonetDB/src/gdk/gdk_utils.mx:1123
2 0x00002aaaab904ad8 in VALclear (v=0x7fffffc51f10)
at /ufs/mk/opensource/MonetDB/src/gdk/gdk_value.mx:123
3 0x00002aaaaaad06bb in interpret_str (stk=Variable
"stk" is not available.
)
at
/ufs/mk/opensource/MonetDB/src/monet/monet_interpreter.mx:247
4 0x00002aaaaaad09a6 in interpret_file (
## Comment 4884
Date: 2006-01-22 19:56:39 +0100
From: @drstmane
Logged In: YES
user_id=572415
The gold tests work fine with `Mtest/py scripts/gold` and by
hand, provided you use the *proper/correct* TSTTRGDIR setting:
========
MonetDB/scripts/gold $ export TSTTRGDIR=$PWD
MonetDB/scripts/gold $ Mserver --dbname=gold load.mil ; echo $?
Monet Database Server V4.11.0
Copyright (c) 1993-2006, CWI. All rights reserved.
Compiled for x86_64-redhat-linux-gnu/64bit with 64bit
OIDs; dynamically linked.
Visit http://monetdb.cwi.nl/ for further information.
MonetDB/scripts/gold $ Mserver --dbname=gold init.mil ; echo $?
Monet Database Server V4.11.0
Copyright (c) 1993-2006, CWI. All rights reserved.
Compiled for x86_64-redhat-linux-gnu/64bit with 64bit
OIDs; dynamically linked.
Visit http://monetdb.cwi.nl/ for further information.
[ 50000 ]
~BeginVariableOutput~
---------------------------------
h t name
oid str type
---------------------------------
[ 20269@0, "damage" ]
[ 39383@0, "damage" ]
~EndVariableOutput~
[ 50000 ]
~BeginVariableOutput~
---------------------------------
h t name
oid str type
---------------------------------
[ 79692@0, "no damage" ]
[ 92777@0, "no damage" ]
~EndVariableOutput~
0
MonetDB/scripts/gold $ Mserver --dbname=gold tst100.mil >
/tmp/gold.out ; echo $?
0
MonetDB/scripts/gold $ wc /tmp/gold.out
974 3356 20522 /tmp/gold.out
========
With the *wrong* TSTTRGDIR setting that you use, I indeed
get a segfault:
========
MonetDB/scripts/gold $ export TSTTRGDIR=""
MonetDB/scripts/gold $ Mserver --dbname=gold load.mil ; echo $?
Monet Database Server V4.11.0
Copyright (c) 1993-2006, CWI. All rights reserved.
Compiled for x86_64-redhat-linux-gnu/64bit with 64bit
OIDs; dynamically linked.
Visit http://monetdb.cwi.nl/ for further information.
!ERROR: ASCIIload: could not open file /format
!ERROR: ascii_io_load: operation failed.
!ERROR: interpret_params: load_data(param 1): evaluation error.
!ERROR: interpret_params: make_persistent(param 1):
evaluation error.
!ERROR: BATSIGcrash: Mserver internal error (Segmentation
fault), please restart.
!ERROR: (One potential cause could be that your disk might
be full...)
Segmentation fault
139
========
I agree that this shouldn't happen, but it's not highest
priority given the time left (~4 hours?) until the (planned)
release...
## Comment 4885
Date: 2006-06-06 21:20:38 +0200
From: @grobian
Logged In: YES
user_id=963970
is this bug fixable by simply checking if TSTTRGDIR (isn't
that a mtest thing by the way) is empty and aborting if so?
## Comment 4886
Date: 2006-11-05 09:23:23 +0100
From: @mlkersten
Logged In: YES
user_id=490798
This error is still open.The result of running the script as
available in the repository gives
Monet Database Server V4.13.1
Copyright (c) 1993-2006, CWI. All rights reserved.
Compiled for x86_64-redhat-linux-gnu/64bit with 64bit
OIDs; dynamically linked.
Visit http://monetdb.cwi.nl/ for further information.
!ERROR: could not open file /format
!ERROR: ascii_io_load: operation failed.
!ERROR: interpret_params: load_data(param 1): evaluation error.
!ERROR: BATSIGcrash: Mserver internal error (Segmentation
fault), please restart.
!ERROR: (One potential cause could be that your disk might
be full...)
Looking into the problem valgrind confirmed that uninitialized
memory is the target of a free, triggering an assertion.
Patching the load script to become
....
make_persistent(load_data(load_format(path+"/format"),
path+"/test100k", 100000));
var fm:=load_format(path+"/format");
fm.print();
make_persistent(load_data(load_format(path+"/format"),
path+"/test100k", 100000));
...
Running this one produces:
...
!ERROR: could not open file /format
!ERROR: ascii_io_load: operation failed.
MonetDB>
Conclusion, the return value of a nested expression may be
bogus if an exception is raised. These problems can
be avoided by setting the return value allways.
## Comment 4887
Date: 2007-01-11 16:47:43 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Originator: NO
Peter suggested the one-liner fix after I found where the problem occured.
Test added.
## Comment 4888
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1411919 at http://sourceforge.net/support/tracker.php?aid=1411919
| Mserver, Crash on Gold demo (when used incorrectly) | https://api.github.com/repos/MonetDB/MonetDB/issues/839/comments | 0 | 2020-11-30T09:01:47Z | 2024-06-27T11:08:40Z | https://github.com/MonetDB/MonetDB/issues/839 | 753,290,746 | 839 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-20 02:37:04 +0100
From: Wouter Alink <<alink>>
To: @njnes
Version: -- development
Last updated: 2006-01-22 09:38:46 +0100
## Comment 4869
Date: 2006-01-20 14:37:04 +0100
From: Wouter Alink <<alink>>
I tried the 'replace()' function.
the XQuery:
replace("a test"," ","")
produces:
ERROR: [pcre_replace]: 1 times inserted nil due to
errors at tuples 0@0.
is this expected behaviour?
## Comment 4870
Date: 2006-01-20 14:42:26 +0100
From: @drstmane
Logged In: YES
user_id=572415
Nope.
your query look fine to me,so it should work.
even if the query were wrong, I'd at least expect a proper
error message.
## Comment 4871
Date: 2006-01-20 15:16:00 +0100
From: @yzchang
Logged In: YES
user_id=341633
But I can not reproduce this error. I get the following:
$echo "fn:replace(\"a test\", \" \", \"\")" | mapiclnt -lxq
"atest"
Wouter: is it possible for you to provide more information?
Thanks!
## Comment 4872
Date: 2006-01-20 16:34:06 +0100
From: Wouter Alink <<alink>>
Logged In: YES
user_id=621590
Maybe I should have noted that I didn't use any
compile-flags. I just used the 'configure_pathfinder'
default options.
Should I have added some flags?
More details: MonetDB CVS (current). PF CVS (current). All
seems to work fine.
## Comment 4873
Date: 2006-01-20 16:43:31 +0100
From: @drstmane
Logged In: YES
user_id=572415
Wouter,
does your system have pcre installed (`pcre-config --version`)?
If not, the replace implementation apparently does not
properly check this and generate a proper error message ...
## Comment 4874
Date: 2006-01-20 16:51:49 +0100
From: @yzchang
Logged In: YES
user_id=341633
I also used the 'current' version to try out the query. My
version is compiled at the beginning of this week, but since
'replace' has been changed for a long time, (theoritically)
it should behave the same.
My version is compiled with the following flags:
--enable-bi1970-01-01 01:01:04 +0100 --enable-debug --disable-optimize
The default options does not use "--enable-debug". I'm
recompiling my version to see if this flag makes difference.
While waiting, I'm just wondering if anyone else can
reproduce this error?
By the way, it also works this way in my version:
MonetDB>xquery("xml", "fn:replace(\"a test\", \" \", \"\")",
false).printf();
<?xml version="1.0" encoding="utf-8"?>
<XQueryResult>
"atest"
</XQueryResult>
MonetDB>
## Comment 4875
Date: 2006-01-20 17:04:39 +0100
From: @yzchang
Logged In: YES
user_id=341633
The implementation of 'fn:replace' checks if 'HAVE_LIBPCRE'
is defined, and
ifndef HAVE_LIBPCRE
GDKerror("pcre_replace() not available as required
version of libpcre was not found by configure.\n");
Maybe this is not enough?
## Comment 4876
Date: 2006-01-20 17:52:14 +0100
From: @grobian
Logged In: YES
user_id=963970
I can confirm (by using dist tarballs even in a very clean
environment):
(pegasus:vtmp/monet/install) fabian% bin/Mserver
Monet Database Server V4.9.3
Copyright (c) 1993-2006, CWI. All rights reserved.
Compiled for x86_64-redhat-linux-gnu/64bit with 64bit
OIDs; dynamically linked.
Visit http://monetdb.cwi.nl/ for further information.
MonetDB>module(pathfinder);
MonetDB>xquery("xml", "fn:replace(\"a test\", \" \", \"\")",
false).printf();
<?xml version="1.0" encoding="utf-8"?>
<XQueryResult>
"atest"
</XQueryResult>
MonetDB>
Our systems have libpcre available.
## Comment 4877
Date: 2006-01-20 19:35:08 +0100
From: @drstmane
Logged In: YES
user_id=572415
Wouter,
(1) could you please confirm that your system is indeed
without pcre (`pcre-config --version`)?
(2) Did you see any other error message than the one line?
If so, which?
"pcre_replace() not available as required version of
libpcre was not found by configure."??
(3) How did you run your query?
Via `MapiClient -lxquery` or via `pf | Mserver` or
via`xquery();` in MIL?
all this information would be more than handy for us to
analyse your bug ...
## Comment 4878
Date: 2006-01-21 16:28:27 +0100
From: Wouter Alink <<alink>>
Logged In: YES
user_id=621590
The command I used:
echo 'replace("a test", " ", "")' | MapiClient -lxq
The error-message was not accompanied by any other error-
message.
I will check the "pcre-config --version" on monday (don't
have access to the machine from home)
## Comment 4879
Date: 2006-01-22 20:01:40 +0100
From: @drstmane
Logged In: YES
user_id=572415
Wouter,
if you test it again, you you please
(1)
get the latest MonetDB/XQuery from the new release branches
"MonetDB_4-10" & "XQuery_0-10" or from
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/.DailyBuilds./
(2)
also try to run your query also via
`xquery("xml","replace(\"a test\", \" \",
\"\")").printf();`
on the Mserver console (with module "pathfinder" loaded)
and sent the complete error messages for both the
"MapiClient" version and the "xquery();" version.
Thanks!
## Comment 4880
Date: 2006-01-22 21:38:46 +0100
From: @drstmane
Logged In: YES
user_id=572415
With the latest version of MonetDB & pathfinder, I get (in
case I compile MonetDB & pathfinder without pcre support):
echo 'replace("a test", " ","")' | MapiClient -lxqERROR:
[pcre_replace]: 1 times inserted nil due to errors at tuples
0@0.
ERROR: [pcre_replace]: first error was:
ERROR: pcre_replace() not available as required version of
libpcre was not found by configure.
ERROR: pcre_replace: operation failed.
probably, Wouter didn't see the complete error message, as
his version didn't have Niels' fix for multi-line error
messages in MapiClient, yet.
done.
## Comment 4881
Date: 2006-01-23 08:35:32 +0100
From: Wouter Alink <<alink>>
Logged In: YES
user_id=621590
As Stefan indeed thought, using:
echo ... | pf > tmp.mil
MapiClient tmp.nil
resulted in the following:
ERROR = !ERROR: [pcre_replace]: 1 times inserted nil due to
errors at tuples 0@0.
!ERROR: [pcre_replace]: first error was:
!ERROR: pcre_replace() not available as required
version of libpcre was not found by configure.
!ERROR: pcre_replace: operation failed.
## Comment 4882
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1410788 at http://sourceforge.net/support/tracker.php?aid=1410788
| XQuery: replace | https://api.github.com/repos/MonetDB/MonetDB/issues/838/comments | 0 | 2020-11-30T09:01:43Z | 2024-06-27T11:08:39Z | https://github.com/MonetDB/MonetDB/issues/838 | 753,290,698 | 838 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-20 11:31:27 +0100
From: @drstmane
To: Jens Teubner <<teubner>>
Version: -- development
Last updated: 2006-06-11 12:05:24 +0200
## Comment 4864
Date: 2006-01-20 11:31:27 +0100
From: @drstmane
The Pathfinder tests tests/XQuery/Tests/typing[12].xq
fail, returning "42.000000" (a doubl/decimal) instead
of "42" (an interger):
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XQuery/typing1.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XQuery/typing2.out.00.html
Jens,
if I get your comment in the test scripts right, this
is indeed a type checking/handling bug in Pathfinder?
Is there a simple way to fix it (i.e., before the
release), or shall we postpone this until after the
release?
[ps: the tests fail since the have been added "long
ago"...]
## Comment 4865
Date: 2006-01-20 11:58:11 +0100
From: Jens Teubner <<teubner>>
Logged In: YES
user_id=731390
I'm not an expert regarding milprint_summer.c. But I guess
the way how types are handled in there (namely solely based
on static types, not on dynamic ones) makes it hard to
implement those test cases correctly.
I think the algebra version correctly handles one of the two
queries right now, fails the same way with the other,
though.
I had added those tests as kind of a reminder that we do
have some minor semantical issues. Maybe we can fix those
one day, but probably not before the release.
## Comment 4866
Date: 2006-06-06 21:18:08 +0200
From: @grobian
Logged In: YES
user_id=963970
Please mark wontfix if bug is not going to be fixed.
## Comment 4867
Date: 2006-06-11 12:05:24 +0200
From: Jens Teubner <<teubner>>
Logged In: YES
user_id=731390
I set this to WONTFIX to make our buglist "clean".
The problem certainly will not be fixed in the
milprint_summer branch. The algebra branch does better
already (Samuel's implementation for built-in functions
should actually do both of them correctly, though I have not
tested).
Both tests are (and have always been) valid and we should
leave them in the test web although they currently are
doomed to fail. I will add a comment into the test scripts.
## Comment 4868
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1410662 at http://sourceforge.net/support/tracker.php?aid=1410662
| PF: tests tests/XQuery/Tests/typing[12].xq fail | https://api.github.com/repos/MonetDB/MonetDB/issues/837/comments | 0 | 2020-11-30T09:01:40Z | 2024-06-28T07:36:47Z | https://github.com/MonetDB/MonetDB/issues/837 | 753,290,659 | 837 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-20 09:37:57 +0100
From: @drstmane
To: @njnes
Version: -- development
Last updated: 2006-04-07 04:35:52 +0200
## Comment 4856
Date: 2006-01-20 09:37:57 +0100
From: @drstmane
MonetDB's `make dist` now fails on SuSE 9ES:
[...]
make[5]: Entering directory
`/localfs/tmp/manegold.GNU.32.32.d.6612/MonetDB/SuSE9ES/src/mapi/clients/perl/Cimpl'
no -perl5 -I../../C -outdir . -o MapiLib.pm.c
/localfs/tmp/manegold.GNU.32.32.d.6612/MonetDB/src/mapi/clients/perl/Cimpl/MapiLib.pm.i
make[5]: no: Command not found
make[5]: *** [MapiLib.pm.c] Error 127
make[5]: Target `distdir' not remade because of errors.
make[5]: Leaving directory
`/localfs/tmp/manegold.GNU.32.32.d.6612/MonetDB/SuSE9ES/src/mapi/clients/perl/Cimpl'
make[4]: *** [distdir] Error 1
[...]
see
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.GNU.32.32.d-SuSE9ES/dist.out
for details
## Comment 4857
Date: 2006-01-20 09:54:27 +0100
From: @drstmane
Logged In: YES
user_id=572415
same holds for sql and on Cygwin:
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.GNU.32.32.s-CYGWIN1.5.12/dist.out
http://monetdb.cwi.nl/testing/projects/monetdb/Current/sql/.GNU.32.32.s-CYGWIN1.5.12/dist.out
http://monetdb.cwi.nl/testing/projects/monetdb/Current/sql/.GNU.32.32.d-SuSE9ES/dist.out
## Comment 4858
Date: 2006-01-20 15:53:57 +0100
From: @drstmane
Logged In: YES
user_id=572415
Problem is that we say:
src/mapi/clients/perl/Makefile.ag:
SUBDIRS = HAVE_PERL_DEVEL?MonetDB-CLI-MapiPP
HAVE_PERL_DEVEL?MonetDB-CLI-MapiLib
HAVE_PERL_DEVEL?MonetDB-CLI HAVE_PERL_DEVEL?DBD
HAVE_PERL_SWIG?Cimpl
autogen (correctly) makes:
src/mapi/clients/perl/Makefile.am:
if HAVE_PERL_DEVEL
HAVE_PERL_DEVEL_1_SUBDIR = MonetDB-CLI-MapiPP
else
HAVE_PERL_DEVEL_1_SUBDIR =
endif
if HAVE_PERL_DEVEL
HAVE_PERL_DEVEL_2_SUBDIR = MonetDB-CLI-MapiLib
else
HAVE_PERL_DEVEL_2_SUBDIR =
endif
if HAVE_PERL_DEVEL
HAVE_PERL_DEVEL_3_SUBDIR = MonetDB-CLI
else
HAVE_PERL_DEVEL_3_SUBDIR =
endif
if HAVE_PERL_DEVEL
HAVE_PERL_DEVEL_4_SUBDIR = DBD
else
HAVE_PERL_DEVEL_4_SUBDIR =
endif
if HAVE_PERL_SWIG
HAVE_PERL_SWIG_5_SUBDIR = Cimpl
else
HAVE_PERL_SWIG_5_SUBDIR =
endif
SUBDIRS = $(HAVE_PERL_DEVEL_1_SUBDIR)
$(HAVE_PERL_DEVEL_2_SUBDIR) $(HAVE_PERL_DEVEL_3_SUBDIR)
$(HAVE_PERL_DEVEL_4_SUBDIR) $(HAVE_PERL_SWIG_5_SUBDIR)
but automake then creates:
src/mapi/clients/perl/Makefile.in:
DIST_SUBDIRS = MonetDB-CLI-MapiPP MonetDB-CLI-MapiLib
MonetDB-CLI DBD \
Cimpl
and then `make dist` tries to make the Cimpl although there
is no SWIG/HAVE_PERL_SWIG ...
maybe/probably, we'd also have autogen explictly set
DIST_SUBDIRS just like SUBDIRS ... ?
... I'll give it a try ...
## Comment 4859
Date: 2006-01-20 16:48:49 +0100
From: @drstmane
Logged In: YES
user_id=572415
Indeed setting DIST_SUBDIRS = $(SUBDIRS) (see attached
patch) would make `make dist` work also if there is no swig.
However, in that case, we would get a "crippled" source
distribution (tarball), which is probably not a good idea.
Hence, it's rather a feature than a bug that `make dist`
failes in case not all vital build tools are present.
... it should give a proper error message, though ...
Moreover, this means that the source distribution also does
not require swig; hence, we should check for it only in case
of the plain CVS version ...
## Comment 4860
Date: 2006-01-20 16:49:31 +0100
From: @drstmane
Logged In: YES
user_id=572415
...forgot attachment...
## Comment 4861
Date: 2006-01-22 19:32:11 +0100
From: @drstmane
Logged In: YES
user_id=572415
As I said before, it's rather a feature than a bug.
leaving it open with lowest priority too keep us reminded
that we should consider producing a more appropriate error
message (if possible) after the release...
## Comment 4862
Date: 2006-04-07 16:35:52 +0200
From: @njnes
Logged In: YES
user_id=43556
make dist now fails directly if one of the required packages
or tools is missing (such as swig and netcdf library)
## Comment 4863
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1410572 at http://sourceforge.net/support/tracker.php?aid=1410572
| `make dist` for MonetDB & sql fails on SuSE 9ES & Cygwin | https://api.github.com/repos/MonetDB/MonetDB/issues/836/comments | 0 | 2020-11-30T09:01:37Z | 2024-06-27T11:08:37Z | https://github.com/MonetDB/MonetDB/issues/836 | 753,290,616 | 836 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-20 09:35:19 +0100
From: @drstmane
To: Fabian Groffen <<fabian>>
Version: -- development
Last updated: 2006-01-20 04:41:43 +0100
## Comment 4845
Date: 2006-01-20 09:35:19 +0100
From: @drstmane
MonetDB's `make rpm` now fails with:
[...]
Making all in java
make[7]: Entering directory
`/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java'
/usr/bin/ant -f ./build.xml
-Dbuilddir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java
-Djardir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java
jar_mapi
Buildfile: ./build.xml
prepare:
compile_mapi:
[echo] Compiling Java Mapi
[javac] Compiling 3 source files to
/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java
jar_mapi:
[jar] Building jar:
/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java/mapi.jar
BUILD SUCCESSFUL
Total time: 2 seconds
make[7]: Leaving directory
`/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java'
[...]
Making all in java
make[7]: Entering directory
`/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/examples/java'
/usr/bin/ant -f ./build.xml
-Dbuilddir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/examples/java
-Djardir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/examples/java
compile
Buildfile: ./build.xml
prepare:
find_driver:
mapi:
prepare:
compile_mapi:
[echo] Compiling Java Mapi
[javac] Compiling 3 source files to
/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/examples/java
jar_mapi:
[jar] Building jar:
/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/examples/java/mapi.jar
compile:
[echo] Compiling Java Mapi Examples
BUILD SUCCESSFUL
Total time: 3 seconds
make[7]: Leaving directory
`/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/examples/java'
[...]
Making install in java
make[7]: Entering directory
`/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java'
/usr/bin/ant -f ./build.xml
-Dbuilddir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java
-Djardir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java
jar_mapi
Buildfile: ./build.xml
prepare:
compile_mapi:
[echo] Compiling Java Mapi
jar_mapi:
BUILD SUCCESSFUL
Total time: 0 seconds
make[8]: Entering directory
`/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java'
/usr/bin/ant -f ./build.xml
-Dbuilddir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java
-Djardir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java
jar_mapi
Buildfile: ./build.xml
prepare:
compile_mapi:
[echo] Compiling Java Mapi
jar_mapi:
BUILD SUCCESSFUL
Total time: 1 second
mkdir -p /var/tmp/MonetDB-root/usr/share/MonetDB/lib
/usr/bin/install -c mapi.jar
/var/tmp/MonetDB-root/usr/share/MonetDB/lib/mapi.jar
make[8]: Nothing to be done for `install-data-am'.
make[8]: Leaving directory
`/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java'
[...]
Making install in java
make[7]: Entering directory
`/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java'
/usr/bin/ant -f ./build.xml
-Dbuilddir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java
-Djardir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java
jar_mapi
Buildfile: ./build.xml
prepare:
compile_mapi:
[echo] Compiling Java Mapi
jar_mapi:
BUILD SUCCESSFUL
Total time: 0 seconds
make[8]: Entering directory
`/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java'
/usr/bin/ant -f ./build.xml
-Dbuilddir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java
-Djardir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java
jar_mapi
Buildfile: ./build.xml
prepare:
compile_mapi:
[echo] Compiling Java Mapi
jar_mapi:
BUILD SUCCESSFUL
Total time: 1 second
mkdir -p /var/tmp/MonetDB-root/usr/share/MonetDB/lib
/usr/bin/install -c mapi.jar
/var/tmp/MonetDB-root/usr/share/MonetDB/lib/mapi.jar
make[8]: Nothing to be done for `install-data-am'.
make[8]: Leaving directory
`/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/clients/java'
[...]
Making install in java
make[7]: Entering directory
`/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/examples/java'
/usr/bin/ant -f ./build.xml
-Dbuilddir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/examples/java
-Djardir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/examples/java
compile
Buildfile: ./build.xml
prepare:
find_driver:
mapi:
prepare:
compile_mapi:
[echo] Compiling Java Mapi
jar_mapi:
compile:
[echo] Compiling Java Mapi Examples
BUILD SUCCESSFUL
Total time: 1 second
make[8]: Entering directory
`/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/examples/java'
/usr/bin/ant -f ./build.xml
-Dbuilddir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/examples/java
-Djardir=/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/examples/java
compile
Buildfile: ./build.xml
prepare:
find_driver:
mapi:
prepare:
compile_mapi:
[echo] Compiling Java Mapi
jar_mapi:
compile:
[echo] Compiling Java Mapi Examples
BUILD SUCCESSFUL
Total time: 1 second
mkdir -p /var/tmp/MonetDB-root/usr/share/MonetDB/Tests
/usr/bin/install -c sample.class
/var/tmp/MonetDB-root/usr/share/MonetDB/Tests/sample.class
/usr/bin/install: cannot stat `sample.class': No such
file or directory
make[8]: *** [install-exec-local-sample_class] Error 1
make[8]: Target `install-exec-am' not remade because of
errors.
make[8]: Nothing to be done for `install-data-am'.
make[8]: Leaving directory
`/export/scratch0/monet/monet.GNU.64.64.d.6612/MonetDB/Fedora4/rpm/BUILD/MonetDB-4.9.3/src/mapi/examples/java'
make[7]: *** [install-am] Error 2
make[7]: Target `install' not remade because of errors.
see
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.GNU.64.64.d-Fedora4/rpm.out
for details.
## Comment 4846
Date: 2006-01-20 09:57:29 +0100
From: @grobian
Logged In: YES
user_id=963970
It seems I'm a bit unfortunate, as I get a lot of errors,
but not this one. What does one need to run `make rpm`?
## Comment 4847
Date: 2006-01-20 10:07:58 +0100
From: @drstmane
Logged In: YES
user_id=572415
(1)
I assigned it to you, first, because it seems java/ant
related, and that's your playground ;-)
have a look at it, and if you think you're not in charge or
cannot fix it, assign it to some else...
(2)
to run rpm, on basically needs a source tarball, then,
rpmbuild runs the usual 'configure; make; make install;'
and complains is not all files listed in the spec file to be
included in the package are present after make install...
## Comment 4848
Date: 2006-01-20 10:14:10 +0100
From: @grobian
Logged In: YES
user_id=963970
will try from scratch. Got Mx errors and "l: command not
found" errors.
## Comment 4849
Date: 2006-01-20 10:31:36 +0100
From: @grobian
Logged In: YES
user_id=963970
Stefan,
I took the nightly sources and compiled them. Problem is
that my workstation reboots itself spontaneously when
compiling gdk_scanselect. I'm affraid helping you will get
a bit difficult today.
## Comment 4850
Date: 2006-01-20 11:35:25 +0100
From: @drstmane
Logged In: YES
user_id=572415
No problem, I'll take-over then...
(Btw: this speeds-up `make rpm` considerable (for testing):
Index: MonetDB.spec.in
===================================================================
RCS file: /cvsroot/monetdb/MonetDB/MonetDB.spec.in,v
retrieving revision 1.27
diff -u -r1.27 MonetDB.spec.in
--- MonetDB.spec.in 16 Dec 2005 12:19:22 -0000 1.27
+++ MonetDB.spec.in 20 Jan 2006 10:34:20 -0000
@@ -143,7 +143,7 @@
./configure \
--enable-assert=no \
- --enable-optimize \
+ --disable-optimize \
--enable-bi1970-01-01 01:00:00 +0100%{bits} \
--with-gcc="%{comp_cc}" \
--with-gxx="%{comp_cxx}" \
;-))
## Comment 4851
Date: 2006-01-20 12:29:13 +0100
From: @grobian
Logged In: YES
user_id=963970
I figured out the same (although on a release test I rather
not use a zillion configure flags :) ), so I will try and
see how for I (or my machine) comes.
## Comment 4852
Date: 2006-01-20 13:58:56 +0100
From: @grobian
Logged In: YES
user_id=963970
Update:
Problem probably found. The make install already fails, in
that it can't find sample.class. This is not that weird, as
the whole sample.java file is not in the dist tar. I added
the file to the EXTRA_DIST in Makefile.ag now. Manually
copying over the file in the dist jar makes it install...
to be continued
## Comment 4853
Date: 2006-01-20 14:27:28 +0100
From: @grobian
Logged In: YES
user_id=963970
ok, the make install issue is fixed in CVS now.
I can't test make rpm, as it seems to compile with
enable-optimize even though I removed the entry from the
spec file...
+ cd MonetDB-4.9.3
+ ./configure --enable-assert=no --enable-optimize
--enable-bi1970-01-01 01:01:04 +0100 '--with-gcc=gcc -m64' --with-gxx=g++
--prefix=/usr --sysconfdir=/etc --localstatedir=/var/lib
--without-php --with-perl
checking build system type... x86_64-redhat-linux-gnu
checking host system type... x86_64-redhat-linux-gnu
checking target system type... x86_64-redhat-linux-gnu
checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... make: ***
wait: Geen kindprocessen. Stop.
make: *** Wachten op onafgeronde taken...
make: *** wait: Geen kindprocessen. Stop.
(pegasus:current/build-pegasus.ins.cwi.nl/MonetDB) fabian%
grep enable-optimize MonetDB.spec
(pegasus:current/build-pegasus.ins.cwi.nl/MonetDB) fabian%
Hence, not closing yet.
## Comment 4854
Date: 2006-01-20 16:41:43 +0100
From: @drstmane
Logged In: YES
user_id=572415
With an additional minot fix in MonetDB.spec.in (see CVS
checkin) it works fine, agian.
## Comment 4855
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1410570 at http://sourceforge.net/support/tracker.php?aid=1410570
| MonetDB's `make rpm` fails | https://api.github.com/repos/MonetDB/MonetDB/issues/835/comments | 0 | 2020-11-30T09:01:33Z | 2024-06-27T11:08:36Z | https://github.com/MonetDB/MonetDB/issues/835 | 753,290,574 | 835 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-20 09:20:04 +0100
From: @drstmane
To: @njnes
Version: -- development
Last updated: 2006-01-20 04:32:59 +0100
## Comment 4840
Date: 2006-01-20 09:20:04 +0100
From: @drstmane
MonetDB's bootstrap now fails on Darwin and Debian (see
below), the only two of our testing platform with
automake < 1.9.2, namely 1.5 (Debian) and 1.6.3 (Darwin).
Do we now require (features of) a newer automake than
1.5, if so which at least? 1.7, 1.8, 1.9 ?
Darwin:
=======
src/Makefile.am:14: variable `BUILT_SOURCES' not defined
src/common/Makefile.am: installing `conf/compile'
src/mapi/clients/Makefile.am:34: variable
`BUILT_SOURCES' not defined
src/mapi/clients/haskell/Makefile.am:14: variable
`BUILT_SOURCES' not defined
src/mapi/clients/java/Makefile.am:37: variable
`BUILT_SOURCES' not defined
src/mapi/clients/php/Cimpl/Makefile.am:32: variable
`BUILT_SOURCES' not defined
src/mapi/clients/php/Makefile.am:15: variable
`BUILT_SOURCES' not defined
src/mapi/clients/python/Cimpl/MonetSQLdb/Makefile.am:55:
variable `BUILT_SOURCES' not defined
src/mapi/clients/tcl/Makefile.am:14: variable
`BUILT_SOURCES' not defined
src/mapi/examples/C/Makefile.am:51: variable
`BUILT_SOURCES' not defined
src/mapi/examples/Makefile.am:20: variable
`BUILT_SOURCES' not defined
src/mapi/examples/java/Makefile.am:41: variable
`BUILT_SOURCES' not defined
src/modules/Makefile.am:24: variable `BUILT_SOURCES'
not defined
src/Makefile.am:14: variable `BUILT_SOURCES' not defined
src/mapi/clients/Makefile.am:34: variable
`BUILT_SOURCES' not defined
src/mapi/clients/haskell/Makefile.am:14: variable
`BUILT_SOURCES' not defined
src/mapi/clients/java/Makefile.am:37: variable
`BUILT_SOURCES' not defined
src/mapi/clients/php/Cimpl/Makefile.am:32: variable
`BUILT_SOURCES' not defined
src/mapi/clients/php/Makefile.am:15: variable
`BUILT_SOURCES' not defined
src/mapi/clients/python/Cimpl/MonetSQLdb/Makefile.am:55:
variable `BUILT_SOURCES' not defined
src/mapi/clients/tcl/Makefile.am:14: variable
`BUILT_SOURCES' not defined
src/mapi/examples/C/Makefile.am:51: variable
`BUILT_SOURCES' not defined
src/mapi/examples/Makefile.am:20: variable
`BUILT_SOURCES' not defined
src/mapi/examples/java/Makefile.am:41: variable
`BUILT_SOURCES' not defined
src/modules/Makefile.am:24: variable `BUILT_SOURCES'
not defined
Debian:
=======
src/Makefile.am:14: variable `BUILT_SOURCES' not defined
automake: configure.in: installing `conf/compile'
src/mapi/clients/Makefile.am:34: variable
`BUILT_SOURCES' not defined
src/mapi/clients/haskell/Makefile.am:14: variable
`BUILT_SOURCES' not defined
src/mapi/clients/java/Makefile.am:37: variable
`BUILT_SOURCES' not defined
src/mapi/clients/php/Cimpl/Makefile.am:32: variable
`BUILT_SOURCES' not defined
src/mapi/clients/php/Makefile.am:15: variable
`BUILT_SOURCES' not defined
src/mapi/clients/python/Cimpl/MonetSQLdb/Makefile.am:55:
variable `BUILT_SOURCES' not defined
src/mapi/clients/tcl/Makefile.am:14: variable
`BUILT_SOURCES' not defined
src/mapi/examples/C/Makefile.am:51: variable
`BUILT_SOURCES' not defined
src/mapi/examples/Makefile.am:20: variable
`BUILT_SOURCES' not defined
src/mapi/examples/java/Makefile.am:41: variable
`BUILT_SOURCES' not defined
src/modules/Makefile.am:24: variable `BUILT_SOURCES'
not defined
## Comment 4841
Date: 2006-01-20 09:29:43 +0100
From: @grobian
Logged In: YES
user_id=963970
Remember that bootstrap is a developers thing IMO, so it
shouldn't be necessary for normal installations (like bison
is for example). You can require developers to have
up-to-date stuff installed... (or not?)
## Comment 4842
Date: 2006-01-20 09:59:25 +0100
From: @drstmane
Logged In: YES
user_id=572415
Sure, agree.
I just wanted
(1) to document the problem.
(2) to know, whether we indeed require a newer automake;
if so, we need to update the check in bootstrap and the
docu accordingly.
## Comment 4843
Date: 2006-01-20 16:32:59 +0100
From: @drstmane
Logged In: YES
user_id=572415
automake <= 1.6.3 (< 1.9.2 ?) seems to require that we
always initialize BUILT_SOURCES,
even if it'\''s empty.
fixed in CVS (buildtools).
## Comment 4844
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1410564 at http://sourceforge.net/support/tracker.php?aid=1410564
| MonetDB's bootstrap fails on Debian & Darwin (automake <1.7) | https://api.github.com/repos/MonetDB/MonetDB/issues/834/comments | 0 | 2020-11-30T09:01:30Z | 2024-06-27T11:08:35Z | https://github.com/MonetDB/MonetDB/issues/834 | 753,290,542 | 834 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-19 12:12:37 +0100
From: Steffen Goeldner <<sgoeldner>>
To: @njnes
Version: -- development
Last updated: 2006-01-22 07:34:14 +0100
## Comment 4830
Date: 2006-01-19 12:12:37 +0100
From: Steffen Goeldner <<sgoeldner>>
select * from env where 1=0 or name = 'host';
crashes the server, whereas
select * from env where name = 'host' and (1=0 or name
= 'host');
gives
[ "config", ... ]
## Comment 4831
Date: 2006-01-20 23:09:40 +0100
From: @njnes
Logged In: YES
user_id=43556
a combination of predicates and or isn't handled by the
compiler yet.
## Comment 4832
Date: 2006-01-20 23:58:46 +0100
From: @njnes
Logged In: YES
user_id=43556
Implemented predicates and or handeling. Second bug was caused
by BATappend not resetting the seqbase (sometimes) with the
first append.
## Comment 4833
Date: 2006-01-21 22:27:33 +0100
From: @njnes
Logged In: YES
user_id=43556
added a test in src/test/bugs/predicate_or-sf-1409824
## Comment 4834
Date: 2006-01-22 19:34:14 +0100
From: @drstmane
Logged In: YES
user_id=572415
test script confirms fix:
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/sql/.mTests103/GNU.64.64.d-Fedora4/src_test_bugs/predicate_or-sf-1409824.out.00.html
## Comment 4835
Date: 2006-01-24 16:32:52 +0100
From: Steffen Goeldner <<sgoeldner>>
Logged In: YES
user_id=1174087
The crash seems to be fixed, but both queries returned
[ "config", ... ]
1h ago.
After the recent gdk_batop.mx fix, the queries returned
[ "host", "localhost" ]
as expected.
## Comment 4836
Date: 2006-01-24 16:41:59 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Another one that we can close?
## Comment 4837
Date: 2006-01-24 17:07:23 +0100
From: Steffen Goeldner <<sgoeldner>>
Logged In: YES
user_id=1174087
AFAICS, it is closed as of 2006-01-22 19:34.
(Did I forgot to reopen it?)
## Comment 4838
Date: 2006-01-24 17:30:52 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
My mistake. I didn't look at the status.
## Comment 4839
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1409824 at http://sourceforge.net/support/tracker.php?aid=1409824
| SQL: Crash: SELECT ... WHERE 1=0 OR ... | https://api.github.com/repos/MonetDB/MonetDB/issues/833/comments | 0 | 2020-11-30T09:01:27Z | 2024-06-27T11:08:34Z | https://github.com/MonetDB/MonetDB/issues/833 | 753,290,500 | 833 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-18 05:07:04 +0100
From: Steffen Goeldner <<sgoeldner>>
To: GDK devs <<bugs-common>>
Version: -- development
Last updated: 2006-01-24 04:39:43 +0100
## Comment 4824
Date: 2006-01-18 17:07:04 +0100
From: Steffen Goeldner <<sgoeldner>>
OS: Windows 2000
Compiler: MS VC++ Toolkit 2003
Parser tools: UnxUtils
Starting the server a 2nd time (with a non-empty var/):
VirtualAlloc(0,1868763136,MEM_COMMIT,PAGE_READWRITE):
failed
GDKvmalloc(1868759048) fail => BBPtrim(enter) usage
[mem=4781012,vm=79953920]
!mallinfo.arena = 2434714
!mallinfo.ordblks = 7317
!mallinfo.smblks = 2338
!mallinfo.hblkhd = 0
!mallinfo.hblks = 0
!mallinfo.usmblks = 2150287
!mallinfo.fsmblks = 30960
!mallinfo.uordblks = 251035
!mallinfo.fordblks = 2432
BBPTRIM_ENTER: memsize=4781012,vmsize=79953920
BBPTRIM: memtarget=1073741824 vmtarget=1073741824
TRIMSCAN: mem=1 vm=1, start=1, limit=1
TRIMSCAN: 8269* 0=tmp_174 (1129)
TRIMSCAN: 8271* 1=tmp_176 (1129)
TRIMSCAN: 8273* 2=tmp_200 (1129)
...
TRIMSCAN: 8321 122=tmp_226 (0)
TRIMSCAN: 8323 123=tmp_227 (0)
TRIMSCAN: end at 1 (size=152)
TRIMSELECT: dirty = 0
TRIMSELECT: candidate=tmp_174 BAT*=5dc59a0
(cnt=1129, mode=4096, refs=0, wait=0,
parent=0, lastused=8269,8269,-2147475379)
TRIMSELECT: keep tmp_174 [13024,12800] bytes
[13024,12800] dirty target(mem=1073741824
vm=1073741824)
TRIMSELECT: candidate=tmp_176 BAT*=5dccdb0
(cnt=1129, mode=4096, refs=0, wait=0,
parent=0, lastused=8271,8271,-2147475377)
TRIMSELECT: keep tmp_176 [18684,18460] bytes
[18684,18460] dirty target(mem=1073741824
vm=1073741824)
TRIMSELECT: candidate=tmp_200 BAT*=5dc8940
...
TRIMSELECT: unload tmp_220 [10468,10244] bytes
[10468,10244] dirty
TRIMSELECT: end
BBPTRIM: lastused batid name
BBPTRIM: 8175 30 tmp_36
BBPTRIM: 8176 31 tmp_37
...
BBPTRIM: 8301 142 tmp_216
BBPTRIM: 8306 144 tmp_220
BBPTRIM_EXIT: memsize=2424852,vmsize=79953920
GDKvmalloc(1868759048) fail => BBPtrim(ready) usage
[mem=3909808,vm=79953920]
!mallinfo.arena = 1849758
!mallinfo.ordblks = 7191
!mallinfo.smblks = 2277
!mallinfo.hblkhd = 0
!mallinfo.hblks = 0
!mallinfo.usmblks = 1292603
!mallinfo.fsmblks = 307944
!mallinfo.uordblks = 237515
!mallinfo.fordblks = 11696
VirtualAlloc(0,1868763136,MEM_COMMIT,PAGE_READWRITE):
failed
!FATAL: GDKvmalloc: failed for 1868759048 bytes
## Comment 4825
Date: 2006-01-20 23:15:07 +0100
From: @njnes
Logged In: YES
user_id=43556
Steffen did you simply start/stop/start an mserver or was
there another reason why there was data in the var directory.
## Comment 4826
Date: 2006-01-24 15:52:27 +0100
From: Steffen Goeldner <<sgoeldner>>
Logged In: YES
user_id=1174087
Yes, simply start/stop/start.
But it seems to be fixed in the meantime.
## Comment 4827
Date: 2006-01-24 16:10:32 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Can we then close this with an "Out of Date"?
## Comment 4828
Date: 2006-01-24 16:25:46 +0100
From: Steffen Goeldner <<sgoeldner>>
Logged In: YES
user_id=1174087
Yes!
## Comment 4829
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1409163 at http://sourceforge.net/support/tracker.php?aid=1409163
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-common@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
| !FATAL: GDKvmalloc: failed for 1868759048 bytes | https://api.github.com/repos/MonetDB/MonetDB/issues/832/comments | 0 | 2020-11-30T09:01:23Z | 2024-06-27T11:08:33Z | https://github.com/MonetDB/MonetDB/issues/832 | 753,290,445 | 832 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-18 04:26:16 +0100
From: @grobian
To: XQuery devs <<bugs-xquery>>
Version: -- development
Last updated: 2006-01-18 04:47:48 +0100
## Comment 4819
Date: 2006-01-18 16:26:16 +0100
From: @grobian
The following query: (used in
invisible_error_messages.SF-1409122.xq)
for $b in
doc("auction.xml")/site/open_auctions/open_auction
where $b/bidder/personref[@person="person18829"] <<
$b/bidder/personref[@person="person10487"]
return <history> {$b/reserve/text()} </history>
produces an error when run:
type error: no variant of function op:node-before
accepts the given argument type(s): node*; node* type
error: maybe you meant:
type error: op:node-before (node, node) as boolean
type error: illegal arguments for function op:node-before
halted in
/ufs/fabian/scratch/monetdb/current/xquery/compiler/semantics/
typecheck.brg (overload), line 1121
When feeding the query to eXist, it doesn't complain on
the syntax. I didn't test any more systems, but I
heard from the original error reporter that on "other
systems" this query also runs. Can any XQuery expert
have a look into it and tell me whether the query is
just correct, or that this is a real bug?
Thanks in advance.
## Comment 4820
Date: 2006-01-18 16:47:48 +0100
From: Jens Teubner <<teubner>>
Logged In: YES
user_id=731390
The compiler is right here.
The two path expressions $b/...[@person = ...] possibly
return more than one node each (or none at all). The '<<'
and '>>' operators are translated into calls to op:node-
before/op:node-after, which only allow singleton nodes as
their arguments.
Pathfinder provides XQuery's static typing feature. This
makes it sometimes a bit more strict regarding to types in
comparison to other implementations. This is, however, a
feature, not a bug.
To make your query correctly typed, wrap both arguments of
'<<' into fn:exactly-one(). This provides the respective
type information at compile time and ensures the correct
quantifier at runtime.
## Comment 4821
Date: 2006-01-18 18:47:46 +0100
From: @grobian
Logged In: YES
user_id=963970
Jens, thanks. As XQuery illiterate, I could have never come
up with such an excellent explanation :)
## Comment 4822
Date: 2006-01-18 19:21:03 +0100
From: @drstmane
Logged In: YES
user_id=572415
Just info:
To avoid further compilation and time-consuming bug-chasing,
I (finally) brought the XMark queries on the XMark web site
in sync with the current Pathfinder version of the queries:
http://monetdb.cwi.nl/xml/Assets/xmlquery.txt
## Comment 4823
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1409135 at http://sourceforge.net/support/tracker.php?aid=1409135
The original assignee of this bug does not have
an account here. Reassigning to the default assignee
for the component, bugs-xquery@monetdb.org.
Previous assignee was nobody@users.sourceforge.net.
| PF: query with "<<" operator doesn't compile | https://api.github.com/repos/MonetDB/MonetDB/issues/831/comments | 0 | 2020-11-30T09:01:20Z | 2024-06-28T07:36:47Z | https://github.com/MonetDB/MonetDB/issues/831 | 753,290,408 | 831 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-18 04:15:00 +0100
From: @grobian
To: @njnes
Version: -- development
Last updated: 2006-01-18 09:54:46 +0100
## Comment 4815
Date: 2006-01-18 16:15:00 +0100
From: @grobian
When using MonetDB/XQuery through MapiClient, error
messages are eaten by the client, resulting in only the
first message to appear.
When using pf on the command line, or JdbcClient, much
more (useful) error messages appear.
I will check a test in that shows this behaviour.
## Comment 4816
Date: 2006-01-18 16:19:30 +0100
From: @grobian
Logged In: YES
user_id=963970
test available as
tests/BugsViaSourgeforce/Tests/invisible_error_messages.SF-1409122.xq
## Comment 4817
Date: 2006-01-18 21:54:46 +0100
From: @njnes
Logged In: YES
user_id=43556
Errors are printed using the clean_print function of mapi.
There was a small bug in the multi line result handling (we
forgot to skip the newline/eos marker).
## Comment 4818
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1409122 at http://sourceforge.net/support/tracker.php?aid=1409122
| PF: MapiClient eats error messages | https://api.github.com/repos/MonetDB/MonetDB/issues/830/comments | 0 | 2020-11-30T09:01:16Z | 2024-06-27T11:08:31Z | https://github.com/MonetDB/MonetDB/issues/830 | 753,290,359 | 830 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-15 02:11:16 +0100
From: @drstmane
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-06-14 12:11:03 +0200
## Comment 4800
Date: 2006-01-15 14:11:16 +0100
From: @drstmane
This bug report summerises the former bug reports
921173 "extremely long and complex query causes
SEGFAULT"
https://sourceforge.net/tracker/index.php?func=detail&aid=921173&group_id=56967&atid=482468
1292727 "Mserver segfault because of
'col_name'"
https://sourceforge.net/tracker/index.php?func=detail&aid=1292727&group_id=56967&atid=482468
1314982 "kill(0) causes SEGFAULT"
https://sourceforge.net/tracker/index.php?func=detail&aid=1314982&group_id=56967&atid=482468
All these bugs seem to be fixed, but the respective
test script cause segfaults on
Fedora Core 4 (32-bit)
Gentoo 1.6.13 (32-bit)
SuSE 9.3 (32-bit)
SuSE 9ES (32-bit)
http://monetdb.cwi.nl/testing/projects/monetdb/Current/sql/.mTests103/GNU.32.32.d-Fedora4/src_test_BugDay_2005-10-06_2.9.3/huge_expression_and_column_name.SF-921173.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/sql/.mTests103/GNU.32.32.d-Gentoo1.6.13/src_test_BugDay_2005-10-06_2.9.3/huge_expression_and_column_name.SF-921173.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/sql/.mTests103/GNU.32.32.d-SuSE9.3/src_test_BugDay_2005-10-06_2.9.3/huge_expression_and_column_name.SF-921173.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/sql/.mTests103/GNU.32.32.d-SuSE9ES/src_test_BugDay_2005-10-06_2.9.3/huge_expression_and_column_name.SF-921173.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-Fedora4/tests_BugsViaSourgeforce/ID.1292727.col_name-segfault.run.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-Gentoo1.6.13/tests_BugsViaSourgeforce/ID.1292727.col_name-segfault.run.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.32.32.d-SuSE9.3/tests_BugsViaSourgeforce/ID.1292727.col_name-segfault.run.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.32.32.d-Fedora4/tests_BugDay_2005-10-06_4.8/kill_0.SF-1314982.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.32.32.d-Gentoo1.6.13/tests_BugDay_2005-10-06_4.8/kill_0.SF-1314982.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.32.32.d-SuSE9.3/tests_BugDay_2005-10-06_4.8/kill_0.SF-1314982.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.32.32.d-SuSE9ES/tests_BugDay_2005-10-06_4.8/kill_0.SF-1314982.err.00.html
Further, also http://sourceforge.net/support/tracker.php?aid=933546 "drop schema crashes server"
https://sourceforge.net/tracker/index.php?func=detail&aid=933546&group_id=56967&atid=482468
might be related...
## Comment 4801
Date: 2006-01-16 22:28:02 +0100
From: @drstmane
Logged In: YES
user_id=572415
The attached debigger trace shows, that at least on Fedora
Core 4 (32-bit0 and Gentoo 1.6.13 (32-bit) the segfault with
test
MonetDB/tests/BugDay_2005-10-06_4.8/Tests/kill_0.SF-1314982.milS
is cause by the fact that in the following snippet of
MonetDB/src/monet/monet_interpreter.mx, the "r =
interpret(stk, lt->yysons[i], res);" "somehow" manages to
set its second parameter "lt->yysons[i]" to NULL, causing
the subsequent "CLEANUP(lt->yysons[i]);" to fail:
========
MonetDB/src/monet/monet_interpreter.mx, revision 1.95, lines
1818-182?
--------
for (i = 0; i < lt->cnt; i++)
if (lt->yysons[i]) {
r = interpret(stk, lt->yysons[i], res);
CLEANUP(lt->yysons[i]);
if (r < 0)
goto xit;
}
========
Further debugging into interpret() has not yet revealed the
spot, where lt->yysons[i] gets (unexpectedly) cleared only
on the 4 platforms mentioned earlier in this bug report.
To cure the sypmtons, and to see whether it also
solved/avoids the segfault on SuSE (32-bit) and with the
other two test scripts, I checked in a "defensive
programming" ((c) Sjoerd ;-)) HACK that guard the
"CLEANUP(lt->yysons[i]);" by an extra "if (lt->yysons[i])".
A propper fix is still to be found; I suspect some "weird"
32/64 bit addressing/initialization/... bug, possibly
somewhere in the (new) yytree implementation...
## Comment 4802
Date: 2006-01-17 12:45:55 +0100
From: @drstmane
Logged In: YES
user_id=572415
Apparenly, my HACK was no solution:
MonetDB/tests/BugDay_2005-10-06_4.8/Tests/kill_0.SF-1314982.milS
now works fine on Fedora Core 4 (32-bit), but triggers a
lt-Mserver:
/var/tmp/monet.GNU.32.32.d.22498/MonetDB/src/common/stream.mx:421:
stream_write: Assertion `s->access == 1' failed.
(some "write after kill" issue?)
on SuSE 9ES, SuSE 9.3, & CygWin 1.5.12 (all 32-bit).
sql/src/test/BugDay_2005-10-06_2.9.3/Tests/huge_expression_and_column_name.SF-921173.sql
(now?) works fine in SuSE 9.3 & SuSE 9ES, but still
segfaults on Gentoo 1.6.13 & Fedora Core 4 (both 32-bit), as
well as on RedHat 4WS (64-bit, on Itanium).
pathfinder/tests/BugsViaSourgeforce/Tests/ID.1292727.col_name-segfault.milS
still fails on Fedora Core 4, Gentoo 1.6.13 & SuSE 9.3 (all
32-bit), (now?) producing a
Mserver:
/local_tmp/manegold.GNU.32.32.d.22498/MonetDB/src/gdk/gdk_utils.mx:1100:
GDKfree: Assertion `(size & 2) == 0' failed.
(some "double-free" issue?).
Any ideas?
## Comment 4803
Date: 2006-01-17 20:24:01 +0100
From: @drstmane
Logged In: YES
user_id=572415
After running for several hours (whole night), gdb's
watchpoints found the following
========
[...]
Program received signal SIGINT, Interrupt.
0x012028a6 in decref (i=9, logical=1, lock=1) at
/ufs/manegold/_/scratch0/Monet/Testing/Current/MonetDB/src/gdk/gdk_bbp.mx:1433
1433 refs = --BBP_lrefs(i);
(gdb) n
1444 b = BBP_cache(i);
(gdb) c
Continuing.
Hardware watchpoint 5: lt->yysons[0]
Old value = (struct YYS *) 0x9948e98
New value = (struct YYS *) 0x9948e00
0x0127f33e in GDKzalloc (size=224) at
/ufs/manegold/_/scratch0/Monet/Testing/Current/MonetDB/src/gdk/gdk_utils.mx:1085
1085 memset(p,0,size);
(gdb) c
Continuing.
Hardware watchpoint 5: lt->yysons[0]
Old value = (struct YYS *) 0x9948e00
New value = (struct YYS *) 0x9940000
0x0127f33e in GDKzalloc (size=224) at
/ufs/manegold/_/scratch0/Monet/Testing/Current/MonetDB/src/gdk/gdk_utils.mx:1085
1085 memset(p,0,size);
(gdb)
Continuing.
Hardware watchpoint 5: lt->yysons[0]
Old value = (struct YYS *) 0x9940000
New value = (struct YYS *) 0x9000000
0x0127f33e in GDKzalloc (size=224) at
/ufs/manegold/_/scratch0/Monet/Testing/Current/MonetDB/src/gdk/gdk_utils.mx:1085
1085 memset(p,0,size);
(gdb)
Continuing.
Hardware watchpoint 5: lt->yysons[0]
Old value = (struct YYS *) 0x9000000
Nng/Current/MonetDB/src/gdk/gdk_utils.mx:1085
1085 memset(p,0,size);
(gdb)
Continuing.
[...]
========
(see attachment for the whole trace).
... the last "Continue" is still busy ...
## Comment 4804
Date: 2006-01-18 09:46:18 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
It looks like this could be a similar problem to bug [
1388697 ] segfault after fork(). The bug there was that
yystree structures got freed as past of bigger yystree
chunks, but some of the structures in the chunks were still
in use.
## Comment 4805
Date: 2006-01-19 09:05:17 +0100
From: @drstmane
Logged In: YES
user_id=572415
With my yesterdays changes (GDKfree now first sets each byte
in a GDKmalloced block to 0xFF before freeing it), a
`kill(0);` now results in
lt-Mserver:
/export/scratch0/monet/monet.GNU.64.64.d.2078/MonetDB/src/common/stream.mx:421:
stream_write: Assertion `s->access == 1' failed.
on ALL platforms:
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/tests_BugDay_2005-10-06_4.8/kill_0.SF-1314982.err.00.html
Additionally, the following tests now also result in
segfaults on (almost) all platforms:
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/tests_BugsViaSourgeforce/ID.914935.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/sql/.mTests103/GNU.64.64.d-Fedora4/src_test_BugDay_2005-10-06_2.9.3/bad_sql_logdir.SF-927371.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/sql/.mTests103/GNU.64.64.d-Fedora4/src_test_bugs/interval_convert_bugs-sf-1274077-1274085.err.00.html
I suppose the cause for these crashes is a (hidden) double
free...
## Comment 4806
Date: 2006-01-19 16:08:57 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
kill(0) has been fixed. The problem there was that the tree
being interpreted was freed by the kill implementation. We
now just disallow killing self or the console.
## Comment 4807
Date: 2006-01-21 00:52:32 +0100
From: @njnes
Logged In: YES
user_id=43556
many of these tests now run, ie can this be closed?
## Comment 4808
Date: 2006-01-21 01:29:45 +0100
From: @drstmane
Logged In: YES
user_id=572415
Let's check the testweb after tonight's test run and decide
then.
## Comment 4809
Date: 2006-01-22 12:14:14 +0100
From: @drstmane
Logged In: YES
user_id=572415
Peter,
could you please have a look at these tests
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/tests_BugsViaSourgeforce/ID.914935.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/sql/.mTests103/GNU.64.64.d-Fedora4/src_test_BugDay_2005-10-06_2.9.3/bad_sql_logdir.SF-927371.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/sql/.mTests103/GNU.64.64.d-Fedora4/src_test_bugs/interval_convert_bugs-sf-1274077-1274085.err.00.html
they segfault since I overwrite the to-be-freed memory in
GDKfree, hence, either the memory is accessed after the
free, or there is a double free...
Thanks,
Stefan
## Comment 4810
Date: 2006-01-22 22:49:33 +0100
From: @drstmane
Logged In: YES
user_id=572415
Most probably related:
several SQL tests trigger a
*** glibc detected *** double free or corruption (out):
0x200000000dc7b930 ***
on Itanium with RedHat Enterprise Linux 4WS and icc:
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/sql/.mTests103/ntv.64.64.d-RedHat4WS/src_benchmarks_arno/08.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/sql/.mTests103/ntv.64.64.d-RedHat4WS/src_benchmarks_arno/39.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/sql/.mTests103/ntv.64.64.d-RedHat4WS/src_benchmarks_tpch/03.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/sql/.mTests103/ntv.64.64.d-RedHat4WS/src_odbc_samples/odbcsample1.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/sql/.mTests103/ntv.64.64.d-RedHat4WS/src_test/news.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/sql/.mTests103/ntv.64.64.d-RedHat4WS/src_test_BugDay_2005-10-06_2.9.3/sesamstr_parser_error.SF-993366.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/sql/.mTests103/ntv.64.64.d-RedHat4WS/src_test_BugDay_2005-11-09_2.8/ORDER_BY_evaluation_error.SF-1023658.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/sql/.mTests103/ntv.64.64.d-RedHat4WS/src_test_BugDay_2005-11-09_2.9.3/delete_x_2.SF-1082915.err.00.html
## Comment 4811
Date: 2006-01-22 22:58:55 +0100
From: @drstmane
Logged In: YES
user_id=572415
In fact, most of the various "randomly scattered" failing
SQL tests (showing either segfaults or bus errors) might be
related:
http://monetdb.cwi.nl/testing/projects/monetdb/Stable/sql/.mTests103/index_short.html
## Comment 4812
Date: 2006-02-07 23:36:28 +0100
From: @njnes
Logged In: YES
user_id=43556
Most of them (if not all) run correctly now, this is mainly
fixed by exiting without cleaning up resources. So there is
still a problem when concurrent cleanup is involved.
## Comment 4813
Date: 2006-06-06 21:15:35 +0200
From: @grobian
Logged In: YES
user_id=963970
niels, what is the status of this bug? Can it be closed?
## Comment 4814
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1406591 at http://sourceforge.net/support/tracker.php?aid=1406591
| several tests cause segfault on the same subset of platforms | https://api.github.com/repos/MonetDB/MonetDB/issues/829/comments | 0 | 2020-11-30T09:01:12Z | 2024-06-28T07:36:46Z | https://github.com/MonetDB/MonetDB/issues/829 | 753,290,298 | 829 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-13 09:35:22 +0100
From: @drstmane
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-01-15 11:42:58 +0100
## Comment 4795
Date: 2006-01-13 21:35:22 +0100
From: @drstmane
In tests
pathfinder/tests/BugTracker/Tests/mposjoin_void_void.SF-1404023.q1{a,b,c}
execute the same query in there different ways:
a)
In Mserver (MIL), first shred the document via
shred_doc(), then run the query via xquery() providing
the shredded documents name to fn:doc().
b)
In Mserver (MIL), run the query via xquery() providing
the documents file name to fn:doc() (on-the-fly shredding).
c)
Run the query directly via `MapiClient -lxquery`
providing the documents file name to fn:doc()
(on-the-fly shredding).
Only the first version works as expected; the other two
result in:
Mserver:
/net/corona.ins.cwi.nl/export/scratch0/manegold/Monet/Testing/Current/pathfinder/runtime/pf_support.mx:1425:
CMDinsertConvert: Assertion `t' failed.
Note:
with test
pathfinder/tests/BugTracker/Tests/mposjoin_void_void.SF-1404023.q2,
on-the-fly shredding (a la version c)) seems to work fine.
Might this be related to the fact that
pathfinder/tests/BugTracker/Tests/mposjoin_void_void.SF-1404023.q1{a,b,c}
shred 2 documents, while
pathfinder/tests/BugTracker/Tests/mposjoin_void_void.SF-1404023.q2
only shreds one?
## Comment 4796
Date: 2006-01-13 22:23:17 +0100
From: @drstmane
Logged In: YES
user_id=572415
Same bug seems to occur in
1204965 "XQuery: Example from W3C fails"
cf.
https://sourceforge.net/tracker/index.php?func=detail&aid=1204965&group_id=56967&atid=482468
## Comment 4797
Date: 2006-01-15 01:10:37 +0100
From: @drstmane
Logged In: YES
user_id=572415
Same bug also seems to occur in
1377006 "XQuery: shredding-on-the-fly"
https://sourceforge.net/tracker/index.php?func=detail&aid=1377006&group_id=56967&atid=482468
## Comment 4798
Date: 2006-01-15 11:42:58 +0100
From: @drstmane
Logged In: YES
user_id=572415
Fixed by back-porting Peter's "tentative" fix for http://sourceforge.net/support/tracker.php?aid=1405188 "PF: shredding documents on-the-fly via fn:doc() fails"
1377006 "XQuery: shredding-on-the-fly"
1204965 "XQuery: Example from W3C fails"
from the XQuery_0-9_update branch to the main trunk.
## Comment 4799
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1405188 at http://sourceforge.net/support/tracker.php?aid=1405188
| PF: shredding documents on-the-fly via fn:doc() fails | https://api.github.com/repos/MonetDB/MonetDB/issues/828/comments | 0 | 2020-11-30T09:01:08Z | 2024-06-27T11:08:29Z | https://github.com/MonetDB/MonetDB/issues/828 | 753,290,249 | 828 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-12 05:27:32 +0100
From: Wouter Alink <<alink>>
To: Stefan Manegold <<Stefan.Manegold>>
Version: -- development
Last updated: 2006-01-13 09:24:14 +0100
## Comment 4788
Date: 2006-01-12 17:27:32 +0100
From: Wouter Alink <<alink>>
The query:
for $i in doc("http://sourceforge.net/support/tracker.php?aid=2.xml")//b
where some $j in distinct-values(doc("http://sourceforge.net/support/tracker.php?aid=1.xml")//aap)
satisfies ends-with(exactly-one($i/@name), $j)
return element { "test" } { $i }
Returns:
ERROR: mposjoin: all BATs in the tail of the third
input BAT (ws_item) must have the same tail type.
I'm not completely sure if it is valid XML..., but the
query:
for $i in doc("http://sourceforge.net/support/tracker.php?aid=2.xml")//b
where some $j in (".jpg",".txt") satisfies
ends-with(exactly-one($i/@name), $j)
return $i
does not return an error...
The documents are:
http://sourceforge.net/support/tracker.php?aid=1.xml:
<ext>
<aap>.jpg</aap>
<aap>.txt</aap>
<aap>.xml</aap>
</ext>
http://sourceforge.net/support/tracker.php?aid=2.xml:
<a>
<b name="iets.jpg"/>
<b name="ietsmeer.jpg"/>
</a>
## Comment 4789
Date: 2006-01-12 17:51:19 +0100
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
Query works for me :)
I would guess that one bat has an oid tail type and another
a void tail :) -- Could you perhaps check this?
## Comment 4790
Date: 2006-01-12 18:26:39 +0100
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
I think I was right: Either mposjoin does not allow
constants or void|void once again triggers a bug. I think
its the latter alternative :)
ws.fetch(ATTR_PROP).print();
ws.fetch(ATTR_PROP).fetch(0).print();
ws.fetch(ATTR_PROP).fetch(1).print();
ws.fetch(ATTR_PROP).fetch(2).print();
ws.fetch(ATTR_FRAG).print();
ws.fetch(ATTR_FRAG).fetch(0).print();
ws.fetch(ATTR_FRAG).fetch(1).print();
ws.fetch(ATTR_FRAG).fetch(2).print();
ws.fetch(PROP_VAL).print();
ws.fetch(PROP_VAL).fetch(0).print();
ws.fetch(PROP_VAL).fetch(1).print();
ws.fetch(PROP_VAL).fetch(2).print();
-------------------------
h t name
void BAT type
-------------------------
[ 0@0, <tmp_127> ]
[ 1@0, <attr_prop0> ]
[ 2@0, <attr_prop1> ]
-----------------
h t name
void oid type
-----------------
-----------------
h t name
void oid type
-----------------
[ 0@0, 0@0 ]
[ 1@0, 1@0 ]
-----------------
h t name
void void type
-----------------
-------------------------
h t name
void BAT type
-------------------------
[ 0@0, <tmp_131> ]
[ 1@0, <tmp_232> ]
[ 2@0, <tmp_303> ]
-----------------
h t name
void oid type
-----------------
[ 1@0 ]
[ 2@0 ]
-------------------------
h t name
void BAT type
-------------------------
[ 0@0, <tmp_133> ]
[ 1@0, <prop_val0> ]
[ 2@0, <prop_val1> ]
-----------------
h t name
void str type
-----------------
-------------------------
h t name
void str type
-------------------------
[ 0@0, "iets.jpg" ]
[ 1@0, "ietsmeer.jpg" ]
-----------------
h t name
void str type
-----------------
## Comment 4791
Date: 2006-01-12 18:30:36 +0100
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
forgot some changes :)
... And one more comment: Loading the files is essential for
generating the bug (just replacing the doc() call by the
elements does not help).
## Comment 4792
Date: 2006-01-13 21:24:14 +0100
From: @drstmane
Logged In: YES
user_id=572415
fixed in CVS:
made mposjoin resistent agains mixing VOID (non-NIL!) and
OID columns.
added test scripts:
pathfinder/tests/BugTracker/Tests/mposjoin_void_void.SF-1404023.*
Note,
the query only works, if the document has been explicitely
pre-shredded; shredding on-the-fly by specifying the
doduments file name with fn:doc() tirggers another bug (see
my next bug report).
## Comment 4793
Date: 2006-01-13 22:19:11 +0100
From: @drstmane
Logged In: YES
user_id=572415
see
https://sourceforge.net/tracker/index.php?func=detail&aid=1405188&group_id=56967&atid=482468
for the new http://sourceforge.net/support/tracker.php?aid=1405188 "PF: shredding documents on-the-fly
via fn:doc() fails".
## Comment 4794
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1404023 at http://sourceforge.net/support/tracker.php?aid=1404023
| mposjoin problems (probably void|void related) | https://api.github.com/repos/MonetDB/MonetDB/issues/827/comments | 0 | 2020-11-30T09:01:04Z | 2024-06-27T11:08:28Z | https://github.com/MonetDB/MonetDB/issues/827 | 753,290,205 | 827 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-10 08:05:56 +0100
From: @njnes
To: @sjoerdmullender
Version: -- development
Last updated: 2006-01-13 05:08:09 +0100
## Comment 4783
Date: 2006-01-10 20:05:56 +0100
From: @njnes
When python is installed in a 32 bit version and
monetdb is compiled with 64 bits configure finds python
and swig.
But the compilation (afcourse) fails. Possible solution
could be to try to compile a python module during
configure and only proceed with python if this succeeds.
## Comment 4784
Date: 2006-01-12 17:05:09 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Please check the fix I checked in and close the bug if it
does indeed fix the bug.
## Comment 4785
Date: 2006-01-13 09:12:04 +0100
From: @njnes
Logged In: YES
user_id=43556
Indeed it works. Could we make it a bit less restrictive, ie
not switch of HAVE_PYTHON but only HAVE_PYTHON_SWIG?.
## Comment 4786
Date: 2006-01-13 17:08:09 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
HAVE_PYTHON is now left alone if we can't compile. Only
HAVE_PYTHON_DEVEL and HAVE_PYTHON_SWIG are turned off.
Closing the bug.
## Comment 4787
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1401931 at http://sourceforge.net/support/tracker.php?aid=1401931
| monetdb vs script language compile options | https://api.github.com/repos/MonetDB/MonetDB/issues/826/comments | 0 | 2020-11-30T09:01:01Z | 2024-06-27T11:08:27Z | https://github.com/MonetDB/MonetDB/issues/826 | 753,290,167 | 826 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-10 01:55:54 +0100
From: thijs <<thijswes>>
To: Stefan Manegold <<Stefan.Manegold>>
Version: -- development
Last updated: 2006-01-11 02:49:08 +0100
## Comment 4777
Date: 2006-01-10 13:55:54 +0100
From: thijs <<thijswes>>
When using enum types to encode strings, some of them
seem to get the same internal representation that is
used for 'not found'. This means that encoding this
particular strings gives the same results as encoding
strings that are not present in the enum table.
Obviously, this may lead to unexpected behaviour.
I tried to compose a minimal example, but did not
succeed in making one where the error occurs. It *does*
occur in the following example based on the
[void,str]-bat in the attached file.
module(enum);
var entities := new(void,str).import("entities.txt");
var enum_entities:=enum_create("enum_entities",entities);
encode(enum_entities,"p").print();
This produces, the expected: [ p ]
encode(enum_entities,"p0").print();
This produces, an empty representation,
as expected (p0 is not in the table): [ ]
encode(enum_entities,"p1").print();
But, this also produces an empty representation,
while "p1" is in the table: [ ]
## Comment 4778
Date: 2006-01-10 14:08:24 +0100
From: @njnes
Logged In: YES
user_id=43556
The file with the bat is missing, could you give it another try?
## Comment 4779
Date: 2006-01-11 02:34:47 +0100
From: @drstmane
Logged In: YES
user_id=572415
test script added as
MonetDB/tests/BugTracker/Tests/empty_enum.SF-1401528.*
## Comment 4780
Date: 2006-01-11 02:41:13 +0100
From: @drstmane
Logged In: YES
user_id=572415
the (extended) test script reveals that encoding fails only
for this one value ("p1"), which happens to by the the 129th
values that is encoded (i.e., assigning IDs starting with 0,
this one get's 128(!)) --- maybe this helps for further
debugging...
## Comment 4781
Date: 2006-01-11 14:49:07 +0100
From: @drstmane
Logged In: YES
user_id=572415
fixed priciple design flaw of the enum encoding in CVS:
(Thanks to Niels for the crucial hit!)
Using internal enum IDs 0..255 (unsigned char) and 0..65535
(unsigned short)
bears the "risk" of hitting 128 (unsigned char) == -128
(signed char) == chr_nil
or 32768 (unsigned short) == -32768 (signed short) ==
shr_nil; which eventually
breaks the enum encoding (especially, since IDs chr_nil and
sht_nil are explicitely
used for non-encoded/non-exiting values...).
To solve the problems, we shift the internal enum IDs to
-127..127 (signed char)
and -32767..32767 (signed short), savely excluding NIL from
the (valid) ID ranges.
NOTE:
Since MonetDB implements NIL as the lowers value with the
domain of its (signed)
data types, using the full ranges of the same type as
unsigned values locates
NIL in the middle of the unsigned ranges, which is bound to
trigger errors unless
special care is taken.
In other words, unsigned data types in MonetDB should (can)
only be used with half
the expected range (cf. OID)!
## Comment 4782
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1401528 at http://sourceforge.net/support/tracker.php?aid=1401528
| 'empty' encoding of enum types | https://api.github.com/repos/MonetDB/MonetDB/issues/825/comments | 0 | 2020-11-30T09:00:58Z | 2024-06-27T11:08:26Z | https://github.com/MonetDB/MonetDB/issues/825 | 753,290,126 | 825 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-09 04:59:14 +0100
From: Jan Rittinger <<jan.rittinger>>
To: Stefan Manegold <<Stefan.Manegold>>
Version: -- development
Last updated: 2006-01-09 09:13:35 +0100
## Comment 4773
Date: 2006-01-09 16:59:14 +0100
From: Jan Rittinger <<jan.rittinger>>
The appended script contains the algebra generated
XMark Query q20 with the inlined element construction.
Inside the element construction (marked with 'BUG
APPEARS HERE') scjoin returns the wrong result as it
possibly switches the seqbases of the void|void
argument bat pre!?
## Comment 4774
Date: 2006-01-09 21:13:35 +0100
From: @drstmane
Logged In: YES
user_id=572415
fixed in CVS (MonetDB);
turned out to be a bug in fetch() on void|void BATs
(took me some time to find out ...)
## Comment 4775
Date: 2006-01-09 21:15:21 +0100
From: @drstmane
Logged In: YES
user_id=572415
for the bookkeeping:
test script added as
pathfinder/tests/BugTracker/Tests/SCJ_with_void_void_BATs.SF-1400473.milS
## Comment 4776
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1400473 at http://sourceforge.net/support/tracker.php?aid=1400473
| scjoin stumbles across void|void bats | https://api.github.com/repos/MonetDB/MonetDB/issues/824/comments | 0 | 2020-11-30T09:00:55Z | 2024-06-27T11:08:25Z | https://github.com/MonetDB/MonetDB/issues/824 | 753,290,083 | 824 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-09 10:24:24 +0100
From: Wouter Alink <<alink>>
To: @sjoerdmullender
Version: -- development
Last updated: 2006-01-13 04:38:33 +0100
## Comment 4767
Date: 2006-01-09 10:24:24 +0100
From: Wouter Alink <<alink>>
When starting Mserver with MIL, SQL and XQuery support,
order does matter:
module(sql_server);
module(pathfinder);
sql_server_start();
pfstart();
mapi_start();
works fine, while if you first load the
pathfinder-module and afterwards the sql_server-module,
the sql_server module does not load correctly.
## Comment 4768
Date: 2006-01-09 10:52:18 +0100
From: @drstmane
Logged In: YES
user_id=572415
Details (Errors, etc.)?
Test script?
Most probably this one is related to
1385247 "Mserver assertion failed with "module(sql_server,
mapi);"
https://sourceforge.net/tracker/index.php?func=detail&aid=1385247&group_id=56967&atid=482468
and
1385152 "start/exit MapiClient with three servers causes
assertion"
https://sourceforge.net/tracker/index.php?func=detail&aid=1385152&group_id=56967&atid=482468
## Comment 4769
Date: 2006-01-09 11:36:10 +0100
From: Wouter Alink <<alink>>
Logged In: YES
user_id=621590
If the --dbinit is used then the following error is displayed:
!ERROR: " ERROR("ds_link(BAT[oid,any::1] l1,
BAT[oid,any::1] r1, ..BAT[oid,any]..) requires an even
number of parameters!\n");"
!
^
! can't help you here, sorry.
If the modules are loaded like on the prompt, the error is:
(after the "module(sql_server);" statement)
!ERROR: " if (bit(n % 2)) {"
! ^
! can't help you here, sorry.
## Comment 4770
Date: 2006-01-13 10:50:03 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
The problem is the definition of the % procs in pf_support.
If you remove those, loading sql_server after pathfinder works.
Of course, pathfinder won't actually work correctly then,
since those definitions are used by the pathfinder code.
## Comment 4771
Date: 2006-01-13 16:38:33 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
I was able to reduce the test to
module(pf_support);
module(mkey);
This is because pf_support defined % (and [%]) as procs and
mkey uses % operators. The MIL parser gets totally confused
in this case.
I have fixed the bug by creating % operators in MonetDB for
dbl (and flt) arguments and removing the procs from pf_support.
A test was added
(pathfinder/tests/BugTracker/Tests/module_order_matters.SF-1400200.*)
One potential problem (which I did not encounter in running
the pathfinder tests) is that we now don't have operators
for lng % dbl and dbl % lng.
## Comment 4772
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1400200 at http://sourceforge.net/support/tracker.php?aid=1400200
| Mserver: module-order does matter | https://api.github.com/repos/MonetDB/MonetDB/issues/823/comments | 0 | 2020-11-30T09:00:51Z | 2024-06-27T11:08:24Z | https://github.com/MonetDB/MonetDB/issues/823 | 753,290,048 | 823 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-09 09:36:32 +0100
From: Wouter Alink <<alink>>
To: @njnes
Version: -- development
Last updated: 2006-01-09 09:55:07 +0100
## Comment 4764
Date: 2006-01-09 09:36:32 +0100
From: Wouter Alink <<alink>>
When starting Mserver with --dbfarm=.... the
log-directory will continue to use the
$MONETDB_PREFIX-path.
If the $MONETDB_PREFIX path is not writable for the
current user then an error appears:
!ERROR:mvc_init: unable to create system tables
!ERROR: logger could not create log directory
/usr/local/var/MonetDB/log/demo/
I could not find any documentation about setting the
log-folder.
## Comment 4765
Date: 2006-01-09 09:55:07 +0100
From: Wouter Alink <<alink>>
Logged In: YES
user_id=621590
In a second attempt, using 'grep', I found an option:
--set sql_logdir=...
I added this to the HowToStart-guide...
## Comment 4766
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1400169 at http://sourceforge.net/support/tracker.php?aid=1400169
| SQL: log-dir != dbfarm-dir | https://api.github.com/repos/MonetDB/MonetDB/issues/822/comments | 0 | 2020-11-30T09:00:48Z | 2024-06-28T07:36:46Z | https://github.com/MonetDB/MonetDB/issues/822 | 753,290,022 | 822 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-06 11:48:31 +0100
From: @drstmane
To: @sjoerdmullender
Version: -- development
Last updated: 2006-01-07 11:30:20 +0100
## Comment 4761
Date: 2006-01-06 23:48:31 +0100
From: @drstmane
Not being familiar with the code, I failed to fix this
properly myself:
icc -no-gcc -DHAVE_CONFIG_H -I.
-I/export/scratch0/monet/monet.ntv.64.32.d.16230/sql/src/odbc/driver
-I../../..
-I/export/scratch0/monet/monet.ntv.64.32.d.16230/sql/src/odbc/driver
-I../include
-I/export/scratch0/monet/monet.ntv.64.32.d.16230/sql/src/odbc/driver/../include
-I../../include
-I/export/scratch0/monet/monet.ntv.64.32.d.16230/sql/src/odbc/driver/../../include
-I/export/scratch0/monet/monet.ntv.64.32.d.16230/MonetDB/.ntv.64.32.d-Fedora4/include/MonetDB
-I/export/scratch0/monet/monet.ntv.64.32.d.16230/MonetDB/.ntv.64.32.d-Fedora4/include/MonetDB/common
-I/export/scratch0/monet/monet.ntv.64.32.d.16230/MonetDB/.ntv.64.32.d-Fedora4/include/MonetDB/gdk
-I/export/scratch0/monet/monet.ntv.64.32.d.16230/MonetDB/.ntv.64.32.d-Fedora4/include/MonetDB/monet
-I/export/scratch0/monet/monet.ntv.64.32.d.16230/MonetDB/.ntv.64.32.d-Fedora4/include/MonetDB/plain
-I/export/scratch0/monet/monet.ntv.64.32.d.16230/MonetDB/.ntv.64.32.d-Fedora4/include/MonetDB/contrib
-I/export/scratch0/monet/monet.ntv.64.32.d.16230/MonetDB/.ntv.64.32.d-Fedora4/include/MonetDB/mapi
-I/export/scratch0/monet/monet.ntv.64.32.d.16230/MonetDB/.ntv.64.32.d-Fedora4/include/MonetDB/C
-DLIBMONETODBC -we140 -wd1418 -c99 -Wall -w2 -mp1 -O3
-restrict -unroll -tpp7 -axWP -we266 -Werror
-wd1418,1419,279,310,981,810,444,193,111,177,171,181,764,269,108,188,1357,102,70,1572
-c
/export/scratch0/monet/monet.ntv.64.32.d.16230/sql/src/odbc/driver/SQLPrepare.c
-KPIC -DPIC -o .libs/libMonetODBC_la-SQLPrepare.o
/export/scratch0/monet/monet.ntv.64.32.d.16230/sql/src/odbc/driver/SQLPrepare.c(58):
error 593: variable "ret" was set but never used
MapiMsg ret;
^
compilation aborted for
/export/scratch0/monet/monet.ntv.64.32.d.16230/sql/src/odbc/driver/SQLPrepare.c
(code 2)
make[4]: *** [libMonetODBC_la-SQLPrepare.lo] Error 1
## Comment 4762
Date: 2006-01-07 11:30:19 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Fixed in CVS.
## Comment 4763
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1398887 at http://sourceforge.net/support/tracker.php?aid=1398887
| SQL: ODBC driver fails to compile with icc | https://api.github.com/repos/MonetDB/MonetDB/issues/821/comments | 0 | 2020-11-30T09:00:45Z | 2024-06-27T11:08:22Z | https://github.com/MonetDB/MonetDB/issues/821 | 753,289,985 | 821 |
[
"MonetDB",
"MonetDB"
] | Date: 2006-01-05 04:46:00 +0100
From: Jan Rittinger <<jan.rittinger>>
To: @njnes
Version: -- development
Last updated: 2006-01-18 09:43:01 +0100
## Comment 4755
Date: 2006-01-05 16:46:00 +0100
From: Jan Rittinger <<jan.rittinger>>
I tried to avoid http://sourceforge.net/support/tracker.php?aid=1385067 and 1334536 to get Mtest
running. However debugmask 10 somehow removes tuples
from a void|void bat. With debugmask 0 everything works
nicely.
## Comment 4756
Date: 2006-01-05 17:43:06 +0100
From: @drstmane
Logged In: YES
user_id=572415
does debugmask 10 give any warnings in that case, or just
silently remove the tuples?
do existing tests suffer from this?
if so, which?
if not, could your supply a (simple) test script to
reproduce the problem?
## Comment 4757
Date: 2006-01-05 20:10:20 +0100
From: Jens Teubner <<teubner>>
Logged In: YES
user_id=731390
Hi Stefan,
I'm not familiar with the problem itself, but Jan has
already attached a test script to this bug report to verify
the problem.
Happy new year!
Jens
## Comment 4758
Date: 2006-01-05 21:00:12 +0100
From: @drstmane
Logged In: YES
user_id=572415
Thanks Jens, Sorry Jan!
... I hadn't seen the attached test script ... I add it to
the test web later ...
## Comment 4759
Date: 2006-01-18 09:43:01 +0100
From: @drstmane
Logged In: YES
user_id=572415
Fixed by Niels void semantic fixes in CVS.
## Comment 4760
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1397840 at http://sourceforge.net/support/tracker.php?aid=1397840
| Mserver debug=10 forgets tuples in void|void bat | https://api.github.com/repos/MonetDB/MonetDB/issues/820/comments | 0 | 2020-11-30T09:00:41Z | 2024-06-27T11:08:21Z | https://github.com/MonetDB/MonetDB/issues/820 | 753,289,924 | 820 |
[
"MonetDB",
"MonetDB"
] | Date: 2005-12-31 12:23:28 +0100
From: @grobian
To: Fabian Groffen <<fabian>>
Version: -- development
Last updated: 2005-12-31 02:47:05 +0100
## Comment 4750
Date: 2005-12-31 12:23:28 +0100
From: @grobian
consider the following scenario:
(pegasus:sql/src/jdbc) fabian% JdbcClient
Connection warning: database specifier not supported on
this server (mserver_sql), protocol version 4
Welcome to the MonetDB interactive JDBC terminal!
Database: MonetDB 4.9.3
Driver: MonetDB Native Driver 1.2 (Blunt+ 20051231)
Type \q to quit, \h for a list of available commands
auto commit mode: on
monetdb-> start transaction;
Operation successful
auto commit mode: off
monetdb-> select 1;
+--------------+
| single_value |
+==============+
| 1 |
+--------------+
1 row
monetdb-> rollback;
Operation successful
auto commit mode: on
Ok, we get what we expect; after the rollback the
auto-commit state is switched on again by the server.
Now if we do the same, but with a rollback:
monetdb-> start transaction;
Operation successful
auto commit mode: off
monetdb-> select 1;
+--------------+
| single_value |
+==============+
| 1 |
+--------------+
1 row
monetdb-> commit;
Operation successful
monetdb->
The auto-commit state is not changed by the server upon
commit, hence the state is still in non-auto-commit,
which is undesirable. One has to explicitly rollback
to get auto-commit enabled again:
monetdb-> rollback;
Operation successful
auto commit mode: on
monetdb->
## Comment 4751
Date: 2005-12-31 14:47:05 +0100
From: @grobian
Logged In: YES
user_id=963970
fixed in CVS.
## Comment 4752
Date: 2006-01-05 17:48:19 +0100
From: @drstmane
Logged In: YES
user_id=572415
Is there a test that "confirms" and monitors that the bug is
and remains fixed?
## Comment 4753
Date: 2006-01-05 17:53:44 +0100
From: @grobian
Logged In: YES
user_id=963970
yes boss, numerous ones. :) In fact all the ones that do:
START TRANSACTION;
...
COMMIT;
START TRANSACTION;
...
ROLLBACK;
as START TRANSACTION is no longer allowed anymore when not
in auto-commit mode.
## Comment 4754
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1394298 at http://sourceforge.net/support/tracker.php?aid=1394298
| SQL: auto commit state is not changed correctly on commit | https://api.github.com/repos/MonetDB/MonetDB/issues/819/comments | 0 | 2020-11-30T09:00:38Z | 2024-06-27T11:08:20Z | https://github.com/MonetDB/MonetDB/issues/819 | 753,289,869 | 819 |
[
"MonetDB",
"MonetDB"
] | Date: 2005-12-30 04:31:10 +0100
From: @drstmane
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-01-15 09:19:36 +0100
## Comment 4746
Date: 2005-12-30 16:31:10 +0100
From: @drstmane
XMark queries 01, 04, 09 produce (slightly) different /
wrong(?) output for the "optimized-modularized" Version,
both with XQuery and MIL modules:
Q01 shows only a minor(?) diff, that might be
acceptable (and hence should be approved):
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XMark_opt_MIL/q01.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XMark_opt_XQ/q01.out.00.html
Q04 produces an empty result:
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XMark_opt_MIL/q04.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XMark_opt_MIL/q04.out.00.html
Q09 seems to produce incomplete results:
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XMark_opt_MIL/q09.out.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XMark_opt_XQ/q09.out.00.html
## Comment 4747
Date: 2006-01-13 19:27:31 +0100
From: @drstmane
Logged In: YES
user_id=572415
I checked in the following changes which seem to solve the
problems.
Peter,
could you please check, whether my changes are OK (i.e., do
not compromise you optimization intensions)?
If everything is fine, please close this bug report.
~/pathfinder $ cvs diff tests/XMark/opt/XQ/Tests/xmark.xq
Index: tests/XMark/opt/XQ/Tests/xmark.xq
===================================================================
RCS file:
/cvsroot/monetdb/pathfinder/tests/XMark/opt/XQ/Tests/xmark.xq,v
retrieving revision 1.2
diff -u -r1.2 xmark.xq
--- tests/XMark/opt/XQ/Tests/xmark.xq 28 Dec 2005 16:15:31
-0000 1.2
+++ tests/XMark/opt/XQ/Tests/xmark.xq 13 Jan 2006 18:22:14
-0000
@@ -5,7 +5,7 @@
2.20371 * $v (: convert Dfl to Euro :)
};
-declare function xmark:q01($doc as xs:string, $id as
xs:string) as xs:string*
+declare function xmark:q01($doc as xs:string, $id as
xs:string) as xs:anyNode*
{
for $b in doc($doc)/descendant::person[@id=$id]
return $b/name/text()
@@ -28,8 +28,8 @@
declare function xmark:q04($doc as xs:string, $person1 as
xs:string, $person2 as xs:string) as xs:anyNode*
{
for $b in doc($doc)/descendant::open_auction
- where some $pr1 in
$b/descendant::personref[@person=$person1],
- $pr2 in
$b/descendant::personref[@person=$person2] satisfies $pr1 <<
$pr2
+ where some $pr1 in
($b/descendant::personref)[@person=$person1],
+ $pr2 in
($b/descendant::personref)[@person=$person2] satisfies $pr1
<< $pr2
return <history> { $b/reserve/text() } </history>
};
@@ -65,8 +65,8 @@
declare function xmark:q09($doc as xs:string) as xs:anyNode*
{
let $auction := doc($doc)
- let $ca := $auction/s/closed_auction
- let $ei := $auction/descendant::item
+ let $ca := $auction/descendant::closed_auction
+ let $ei := $auction/descendant::europe/item
for $p in $auction/descendant::person
let $a :=
for $t in $ca
## Comment 4748
Date: 2006-01-15 21:19:36 +0100
From: @peterboncz
Logged In: YES
user_id=591107
I have some additional optimizations, but what you did was
included in that, yes.
## Comment 4749
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1393682 at http://sourceforge.net/support/tracker.php?aid=1393682
| PF: tests q0{1,4,9} fail in tests/XMark/opt/{MIL,XQ} | https://api.github.com/repos/MonetDB/MonetDB/issues/818/comments | 0 | 2020-11-30T09:00:35Z | 2024-06-27T11:08:19Z | https://github.com/MonetDB/MonetDB/issues/818 | 753,289,821 | 818 |
[
"MonetDB",
"MonetDB"
] | Date: 2005-12-23 10:11:26 +0100
From: @sjoerdmullender
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2005-12-23 04:20:22 +0100
## Comment 4739
Date: 2005-12-23 10:11:26 +0100
From: @sjoerdmullender
When I start
Mserver --dbinit 'module(mapi); mapi_start();'
I consistently get a segfault. After many hours spent
in gdb and valgrind (totally useless in this case), I
found the problem, although I don't have a fix yet.
The problem is a combination of fork() and the new way
of allocating yystree structures.
The command line argument results in a parse tree.
When this parse tree is executed, CMDfork() is called
to execute the MIL command fork() (inside
mapi_start()). CMDfork() forks off a new thread and
creates a client context, and saves the first argument
in the client context (c->tree = lt->yysons[0];
lt->yysons[0] = NULL;). Later the parse tree is freed
by Myyfree() called from interpret_str() which is the
function that executed the command line script.
Myyfree() recursively goes through the parse tree and
frees all nodes. Of course, it will hit the node
corresponding with the call to fork(), but it will skip
the first parameter because that one was zapped.
However, and here we get to the new way in which the
nodes are allocated, when freeing a yystree structure,
there are two possibilities. Either it was the first
such structure in a chunk (the malloced bit in set), or
it wasn't (the bit is cleared). If the malloced bit is
set, the structure is freed, and when it isn't it's
left alone. But, and this is the killer, when the
structure is freed, it takes along other structures in
the same chunk! And it so happened that the head of
the saved parse tree was freed together with some other
yystree structure, so that when the argument of fork()
is later executed, the pointer to the top was actually
pointing into free memory and was being overwritten by
a call to GDKmalloc() (and malloc()) while being
interpreted.
I see two possibilities to solve this:
- Don't free parse trees, especially when there are
calls to fork();
- Make a deep copy in CMDfork of the subtree (and let
the original subtree get freed together with the rest
in Myyfree()).
Another possibility is to rethink the way parse trees
are freed and only free chunks when all the yystree
nodes on them have been freed. This is a far more
complex solution, I would think.
Peter, what is your opinion?
## Comment 4740
Date: 2005-12-23 13:30:07 +0100
From: @sjoerdmullender
Logged In: YES
user_id=43607
Attached is a tentative fix. It's a fix because I don't get
the crash anymore and the MonetDB tests run no worse than
without. It's tentative because I want to hear opnions.
The fix introduces a new function Myycopy which calls yycopy
with the correct parameters to make a deep copy of a parse
tree. This function is then called from CMDfork to copy the
tree.
I didn't use Myylogout, even though that is very similar to
the new Myycopy, since that calls yypostprocess on the sub
tree and that gave a failed assertion.
## Comment 4741
Date: 2005-12-23 16:05:19 +0100
From: @peterboncz
Logged In: YES
user_id=591107
Hi Sjoerd,
Nice find. It may explain a lot of the strange and
hard-to-repeat segmentation violations when quitting Mserver
in the xquery test web (which also forks).
These strange segmentation violations started to appear
after th yytree change.
Your solution with a deep copy is OK.
An alternative would be a hack in yycopy to detect fork
calls and treat them in a special way.
But the thing inside the fork is typically a single function
call, hence small. So deep copying it is no problem.
Peter
## Comment 4742
Date: 2005-12-23 16:16:28 +0100
From: @drstmane
Logged In: YES
user_id=572415
Sjoerd,
since Peter agrees, and I also hope that this might some
several of the "strange" segfault we get since the yytree
changes, could you please check in your changes?
Then, TestWeb will tell us more ... and in case thing
unexpectedly get "worse", we can easily undo the (small)
chnages...
Thank you very much for detecting, analysing and fixing this
problem!
Stefan
## Comment 4743
Date: 2005-12-23 16:26:24 +0100
From: @drstmane
Logged In: YES
user_id=572415
Oops, that was a quick close --- I haven't even seen
Sjoerd's check-in, yet ... (not even mentioning confirmation
by testing results ...)
... sorry, sitting in Germany, I become (even) more German,
again ... ;-))
## Comment 4744
Date: 2005-12-23 18:38:19 +0100
From: @drstmane
Logged In: YES
user_id=572415
just for info:
A first quick test seems to confirm, that Sjoerd's fix
indeed also fixes the "misterious" segfault with
Pathfinder/XQuery tests.
Thanks, again!
## Comment 4745
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1388697 at http://sourceforge.net/support/tracker.php?aid=1388697
| segfault after fork() | https://api.github.com/repos/MonetDB/MonetDB/issues/817/comments | 0 | 2020-11-30T09:00:31Z | 2024-06-27T11:08:18Z | https://github.com/MonetDB/MonetDB/issues/817 | 753,289,783 | 817 |
[
"MonetDB",
"MonetDB"
] | Date: 2005-12-22 06:50:50 +0100
From: Jan Rittinger <<jan.rittinger>>
To: Jan Rittinger <<jan.rittinger>>
Version: -- development
Last updated: 2006-01-18 01:40:07 +0100
## Comment 4729
Date: 2005-12-22 18:50:50 +0100
From: Jan Rittinger <<jan.rittinger>>
The following small script generates the error:
!ERROR: XML Generation: (null) BAT does not have a 0 head.
!ERROR:
!ERROR: xquery_print_result_main: operation failed.
module(pathfinder);
var ws := create_ws ();
var a0003 := new (lng, void).key (true).reverse
().seqbase (0@0);
var a0004 := new (dbl, void).key (true).reverse
().seqbase (0@0);
var a0005 := new (dbl, void).key (true).reverse
().seqbase (0@0);
var a0006 := new (str, void).key (true).reverse
().seqbase (0@0);
var a0012 := new (void, oid).seqbase (0@0);
var a0013 := new (void, int).seqbase (0@0);
a0013.insert (nil, INT);
a0013.insert (nil, DEC);
a0005.insert (nil, dbl(2.3));
a0003.insert (nil, lng(1));
a0012.insert (nil, 0@0);
a0012.insert (nil, 0@0);
a0012 := a0012.reverse ().mark (0@0).reverse ();
a0013 := a0013.reverse ().mark (0@0).reverse ();
print(a0005);
print(a0013);
print(a0012);
print_result ("xml", ws, a0012, a0013, a0003, a0004,
a0005, a0006);
## Comment 4730
Date: 2005-12-29 14:02:36 +0100
From: @drstmane
Logged In: YES
user_id=572415
added test script in
pathfinder/tests/BugTracker/Tests/serialization_of_decimal.SF-1388225.*
With the extended error messaging for "idxTV" in
runtime/serialize.mx, I now get:
!ERROR: XML Generation: (null) BAT is empty.
!ERROR:
operation failed.
Jan (R),
is the test script indeed "consistent"?
## Comment 4731
Date: 2006-01-02 09:26:34 +0100
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
Patched the correct expected result - testscript seems
correct to me.
However somebody changed the serialization and now just
discards decimal results!!!
## Comment 4732
Date: 2006-01-14 15:47:03 +0100
From: @drstmane
Logged In: YES
user_id=572415
In the testweb, we still/again get
!ERROR: XML Generation: (null) BAT is empty.
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/ntv.64.32.d-Fedora4/tests_BugTracker/serialization_of_decimal.SF-1388225.out.00.html
## Comment 4733
Date: 2006-01-14 21:04:13 +0100
From: @drstmane
Logged In: YES
user_id=572415
assigned to Peter, since his changes of Dec 06 2005 seem to
have removed the code for serializing decimals (decVAL) ---
deliberately or accidently?
===================================================================
2005/12/06 - boncz: compiler/compile.c,1.58
compiler/mil/milprint_summer.c,1.143
runtime/pf_support.mx,1.73
runtime/serialize.mx,1.40
compile.c:
statement include "algopt_icol.h" leads to errors and
was deleted (Jan??)
milprint_summer.c:
some soap-related check-ins part of what Jennie
checke-ed in yesterday (but forgot about these)
pf_support.mx/serialize.mx:
enhancements to print_result:
- made it faster by taking a perf-critical look at the
idxTV set of primitives
- added a second signature that takes an additional
loop[oid,any] and iter[void,oid] bats.
this function can print multiple iterations as
separate trees.
We need it for soap and it may also be handy for
debugging purposes.
- added a 'noheader' xml-mode modifier that makes pf omit
<?xml version="1.0" encoding="utf-8"?> header
- added a 'root-XYZ' xml-mode modifier (that should
always be last in the mode string).
It uses XYZ as the root element name (instead of the
default XQueryResult)
===================================================================
http://cvs.sourceforge.net/viewcvs.py/monetdb/pathfinder/runtime/serialize.mx?r1=1.39&r2=1.40
`cvs update -p -r1.1 runtime/serialize.mx | grep -B7 -A4
decVAL`
`cvs update -p -r1.39 runtime/serialize.mx | grep -B7 -A4
decVAL`
`cvs update -p -r1.40 runtime/serialize.mx | grep -B7 -A4
decVAL`
`cvs update -p -rHEAD runtime/serialize.mx | grep -B7 -A4
decVAL`
## Comment 4734
Date: 2006-01-15 21:18:45 +0100
From: @peterboncz
Logged In: YES
user_id=591107
Hi Jan,
yeah. decimal support in serialize is kind of terminated;
that is obvious if you look at the implementation of
xquery_print_result_main
milprint_summer uses the same value container for dbl and
dec (both variables refer to the same bat). Until the
point that Monet implements a usable decimal, we are stuck
with using dbl, and then one can just put them together,
not?
well, that is the current situation, at least.
## Comment 4735
Date: 2006-01-15 21:39:23 +0100
From: @drstmane
Logged In: YES
user_id=572415
Let's keep this one "open" to remind us, that there is still
some work to do to fix this, either by reviving the
serialization of decimals, or by properly mapping it to dbl...
## Comment 4736
Date: 2006-01-17 13:58:52 +0100
From: @drstmane
Logged In: YES
user_id=572415
Jan,
since Niels made the MIL query work, is there a XQ query
that could trigger/check the decimal serialization?
Thanks,
Stefan
## Comment 4737
Date: 2006-01-18 13:40:07 +0100
From: Jan Rittinger <<jan.rittinger>>
Logged In: YES
user_id=993208
Like Peters change in milprint_summer.c Niels fix in
milgen.brg is a workaround that avoids the error message.
The consequence is that the algebra translation has to
translate decimals as doubles as well...
## Comment 4738
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1388225 at http://sourceforge.net/support/tracker.php?aid=1388225
| serialization of decimal does not work | https://api.github.com/repos/MonetDB/MonetDB/issues/816/comments | 0 | 2020-11-30T09:00:28Z | 2024-06-27T11:08:17Z | https://github.com/MonetDB/MonetDB/issues/816 | 753,289,755 | 816 |
[
"MonetDB",
"MonetDB"
] | Date: 2005-12-21 09:36:03 +0100
From: @drstmane
To: Stefan Manegold <<Stefan.Manegold>>
Version: -- development
Last updated: 2005-12-27 07:30:29 +0100
## Comment 4724
Date: 2005-12-21 09:36:03 +0100
From: @drstmane
since Wednesday Dec 21 2005, test
tests/from_Twente/moa/Tests/list.milM fails with
!WARNING: BATpropcheck: BAT tmp_211 was incorrectly
marked sorted!
!WARNING: BATpropcheck: BAT tmp_211 remains marked
radix-clustered on 32 bits; not checked!
!WARNING: BATpropcheck: BAT tmp_211 was incorrectly
marked sorted!
!WARNING: BATpropcheck: BAT tmp_211 remains marked
radix-clustered on 32 bits; not checked!
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/tests_from_Twente_moa/list.out.00.html
## Comment 4725
Date: 2005-12-25 18:41:23 +0100
From: @drstmane
Logged In: YES
user_id=572415
This occurs since Dec 21 2005, might hence be related to
changes made on Dec 20 2005.
## Comment 4726
Date: 2005-12-27 14:50:15 +0100
From: @drstmane
Logged In: YES
user_id=572415
in fact, this appears to be a buggy sunion of two
[void,void] BATs instead of "just" a property problem:
_MMjoin_90.print();
-----------------
h t name
void void type
-----------------
[ 0@0, 0@0 ]
[ 1@0, 1@0 ]
[ 2@0, 2@0 ]
_MMjoin_92.print();
-----------------
t h name
void void type
-----------------
[ 0@0, 0@0 ]
[ 1@0, 1@0 ]
[ 2@0, 2@0 ]
var _MMsunion_93 := sunion(_MMjoin_90,_MMjoin_92);
_MMsunion_93.print();
! wrong !
-----------------
h t name
oid oid type
-----------------
[ 2@0, 2@0 ]
[ 0@0, 0@0 ]
[ 0@0, 0@0 ]
[ 0@0, 0@0 ]
[ 1@0, 1@0 ]
[ 2@0, 2@0 ]
! wrong !
! correct/expected !
-----------------
h t name
oid oid type
-----------------
[ 0@0, 0@0 ]
[ 1@0, 1@0 ]
[ 2@0, 2@0 ]
! correct/expected !
## Comment 4727
Date: 2005-12-27 19:30:29 +0100
From: @drstmane
Logged In: YES
user_id=572415
Actually, this test uses its own local "sunion" PROC that
merely concatenates two BATs by inserting both into a new
empty BAT; in this case, the new empty BAT was [void,void],
which triggered the known [void,void] insert/append bug
(cf.
https://sourceforge.net/tracker/index.php?func=detail&aid=1385067&group_id=56967&atid=482468
https://sourceforge.net/tracker/index.php?func=detail&aid=1353017&group_id=56967&atid=482468
https://sourceforge.net/tracker/index.php?func=detail&aid=1340615&group_id=56967&atid=482468
https://sourceforge.net/tracker/index.php?func=detail&aid=1334536&group_id=56967&atid=482468
)
Since the bug is tested elsewhere, we avoid it here by
making the new empty BAT [oid,oid], and hence close this report.
## Comment 4728
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1386883 at http://sourceforge.net/support/tracker.php?aid=1386883
This bug was marked DUPLICATE in the database it was moved from.
Changing resolution to "MOVED"
| buggy sunion of two [void,void] BATs | https://api.github.com/repos/MonetDB/MonetDB/issues/815/comments | 0 | 2020-11-30T09:00:25Z | 2024-06-28T07:36:46Z | https://github.com/MonetDB/MonetDB/issues/815 | 753,289,717 | 815 |
[
"MonetDB",
"MonetDB"
] | Date: 2005-12-21 12:35:15 +0100
From: @drstmane
To: Stefan Manegold <<Stefan.Manegold>>
Version: -- development
Last updated: 2005-12-25 06:43:51 +0100
## Comment 4721
Date: 2005-12-21 00:35:15 +0100
From: @drstmane
Pathfinder tests that run XQuery queries via MapiClient
-lxquery randomly cause a segfault when Mserver is
shutdown (via a "quit();" on the MIL console) AFTER the
test has run SUCCESSFULLY...
This happens (at least within Mtest.py) for several
(but different) tests each night. However, the
randomness admittedly makes it difficult to debug and
fix...
## Comment 4722
Date: 2005-12-25 18:45:29 +0100
From: @drstmane
Logged In: YES
user_id=572415
Seems to be fixed by Sjoerd's recent fixes for http://sourceforge.net/support/tracker.php?aid=1388697
"segfault after fork()"; cf.
https://sourceforge.net/tracker/index.php?func=detail&aid=1388697&group_id=56967&atid=482468
## Comment 4723
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1386611 at http://sourceforge.net/support/tracker.php?aid=1386611
| PF: MapiClient tests randomly cause segfault in Mserver | https://api.github.com/repos/MonetDB/MonetDB/issues/814/comments | 0 | 2020-11-30T09:00:22Z | 2024-06-27T11:08:15Z | https://github.com/MonetDB/MonetDB/issues/814 | 753,289,685 | 814 |
[
"MonetDB",
"MonetDB"
] | Date: 2005-12-21 12:27:36 +0100
From: @drstmane
To: Stefan Manegold <<Stefan.Manegold>>
Version: -- development
Last updated: 2006-01-13 11:30:30 +0100
## Comment 4717
Date: 2005-12-21 00:27:36 +0100
From: @drstmane
test pathfinder/tests/XQuery/Tests/orderby1.xq fails with:
WARNING: BATpropcheck: BAT tmp_344 was incorrectly
marked sorted!
http://monetdb.cwi.nl/testing/projects/monetdb/Current/pathfinder/.mTests103/GNU.64.64.d-Fedora4/tests_XQuery/orderby1.err.00.html
## Comment 4718
Date: 2005-12-25 18:47:50 +0100
From: @drstmane
Logged In: YES
user_id=572415
Seems to occur since Oct 05 2005, might hence be related to
changes made of Oct 04 2005.
## Comment 4719
Date: 2006-01-13 23:30:30 +0100
From: @drstmane
Logged In: YES
user_id=572415
fixed sort-property propagation in bat_arith
## Comment 4720
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1386607 at http://sourceforge.net/support/tracker.php?aid=1386607
| PF: property error in tests/XQuery/Tests/orderby1.xq | https://api.github.com/repos/MonetDB/MonetDB/issues/813/comments | 0 | 2020-11-30T09:00:19Z | 2024-06-27T11:08:14Z | https://github.com/MonetDB/MonetDB/issues/813 | 753,289,652 | 813 |
[
"MonetDB",
"MonetDB"
] | Date: 2005-12-21 12:00:41 +0100
From: @drstmane
To: Fabian Groffen <<fabian>>
Version: -- development
Last updated: 2006-01-19 04:09:32 +0100
## Comment 4708
Date: 2005-12-21 00:00:41 +0100
From: @drstmane
The tests in MonetDB/src/mapi/examples/java/Tests/ fail
with
Exception in thread "main"
java.lang.NoClassDefFoundError: sample
cf. e.g.
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/src_mapi_examples_java/sample0.err.00.html
http://monetdb.cwi.nl/testing/projects/monetdb/Current/MonetDB/.mTests103/GNU.64.64.d-Fedora4/src_mapi_examples_java/sample1.err.00.html
## Comment 4709
Date: 2005-12-25 18:50:07 +0100
From: @drstmane
Logged In: YES
user_id=572415
Seems to occur since Dec 11 2005, might hence be related to
changes made on Dec 10 2005.
## Comment 4710
Date: 2005-12-27 19:51:02 +0100
From: @grobian
Logged In: YES
user_id=963970
This is a problem in building and running the examples.
Will try and see if I can fix it.
## Comment 4711
Date: 2005-12-27 20:10:14 +0100
From: @grobian
Logged In: YES
user_id=963970
the problem is somewhat trickier.
compilation of the samples is currently disabled because
it's hard to find the mapi.jar file...
## Comment 4712
Date: 2005-12-29 20:32:14 +0100
From: @drstmane
Logged In: YES
user_id=572415
Well, during `make` (and hence during `make check`), the
mapi.jar is made/found in
$MONETDB_BUILD/src/mapi/clients/java/mapi.jar
After `make install` (and hence during `Mtest`), the
mapi.jar is installed/found in
$MONETDB_PREFIX/share/MonetDB/lib/mapi.jar
A test script could assume the latter as default (for
`Mtest`) and we could add the former to CLASSPATH in
RunMtest.in (for `make check`).
Would that help?
## Comment 4713
Date: 2005-12-29 21:03:40 +0100
From: @grobian
Logged In: YES
user_id=963970
autogen must provide the build.xml file with the right
location for the mapi.jar file. Because both jar and sample
applications are in different trees, I have no idea how to
handle that a bit elegant. I can hardcode some paths, but
that will mean the build system gets screwed up more or less.
## Comment 4714
Date: 2006-01-15 12:25:12 +0100
From: @grobian
Logged In: YES
user_id=963970
should fix this...
## Comment 4715
Date: 2006-01-19 16:09:32 +0100
From: @grobian
Logged In: YES
user_id=963970
should be fixed now.
## Comment 4716
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1386596 at http://sourceforge.net/support/tracker.php?aid=1386596
| mapi java tests fails with java exception | https://api.github.com/repos/MonetDB/MonetDB/issues/812/comments | 0 | 2020-11-30T09:00:16Z | 2024-06-27T11:08:13Z | https://github.com/MonetDB/MonetDB/issues/812 | 753,289,614 | 812 |
[
"MonetDB",
"MonetDB"
] | Date: 2005-12-20 08:29:51 +0100
From: @mlkersten
To: Peter Boncz <<boncz>>
Version: -- development
Last updated: 2006-01-15 09:11:26 +0100
## Comment 4703
Date: 2005-12-20 20:29:51 +0100
From: @mlkersten
./suite_02/Tests/test_00.valgrind.out:==25088== ERROR
SUMMARY: 6 errors from 1 contexts (suppressed: 13 from 5)
==26351== Syscall param write(buf) points to
uninitialised byte(s)
==26351== at 0x1307D15B: (within
/lib64/libpthread-2.3.5.so)
==26351== by 0x129F747F: GDKsave (gdk_storage.mx:281)
==26351== by 0x129F8ED4: BATsave (gdk_storage.mx:531)
==26351== by 0x12983311: BBPsync (gdk_bbp.mx:2235)
==26351== by 0x12978169: TMcommit (gdk_tm.mx:85)
==26351== by 0x1247E3B8: global_commit (trans.mx:111)
==26351== by 0x124788B3:
global_commit_unpack1567181596 (trans.glue.c:63)
==26351== by 0x11C3D7C4: interpret
(monet_interpreter.mx:1169)
==26351== by 0x11C45491: interpret_seqblock
(monet_interpreter.mx:1817)
==26351== by 0x11C3CAF9: interpret
(monet_interpreter.mx:766)
==26351== by 0x11C47399: interpret_str
(monet_interpreter.mx:245)
==26351== by 0x11C47695: interpret_file
(monet_interpreter.mx:290)
==26351== Address 0x1D1285DA is 4162 bytes inside a
block of size 124108 alloc'd
==26351== at 0x11B1FE96: malloc
(vg_replace_malloc.c:149)
==26351== by 0x129DA289: GDKmallocmax
(gdk_utils.mx:1056)
==26351== by 0x12986560: HEAPalloc (gdk_heap.mx:125)
==26351== by 0x129DD6B0: strHeap (gdk_atoms.mx:1672)
==26351== by 0x129DB6A8: ATOMheap (gdk_atoms.mx:904)
==26351== by 0x129FBBC6: BATnewstorage (gdk_bat.mx:217)
==26351== by 0x129FBE75: BATnew (gdk_bat.mx:250)
==26351== by 0x1D17C7F9: WISCgenerator (wisc.mx:215)
==26351== by 0x1D17C474:
WISCgenerator_unpack963085959 (wisc.glue.c:42)
==26351== by 0x11C3D7C4: interpret
(monet_interpreter.mx:1169)
==26351== by 0x11C45491: interpret_seqblock
(monet_interpreter.mx:1817)
==26351== by 0x11C3CAF9: interpret
(monet_interpreter.mx:766)
## Comment 4704
Date: 2005-12-21 14:11:30 +0100
From: @mlkersten
Logged In: YES
user_id=490798
All valgrind tests were executed with the following simple
script (milS,milM)
for f in `ls tst???.milS`
do
echo "---->" $f
valgrind mserver -d0 $f </dev/null >`basename $f
.mal`.valgrind.out 2>&1done
## Comment 4705
Date: 2006-01-15 16:41:38 +0100
From: @mlkersten
Logged In: YES
user_id=490798
Check urgency of this bug and fix
## Comment 4706
Date: 2006-01-15 21:11:26 +0100
From: @peterboncz
Logged In: YES
user_id=591107
sorry, but this is a warning about the system libraries
## Comment 4707
Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <<sfimport>>
This bug was previously known as tracker item 1386443 at http://sourceforge.net/support/tracker.php?aid=1386443
| Mserver: Valgrind detected error 8 | https://api.github.com/repos/MonetDB/MonetDB/issues/811/comments | 0 | 2020-11-30T09:00:13Z | 2024-06-28T07:36:46Z | https://github.com/MonetDB/MonetDB/issues/811 | 753,289,564 | 811 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.