id stringlengths 22 25 | commit_message stringlengths 137 6.96k | diffs listlengths 0 63 |
|---|---|---|
derby-DERBY-3812-2301e09b | DERBY-3812 - fix test regression in
NetworkServerMBeanTest.testAttributeDrdaStreamOutBufferSize caused by checkin
of converted test OutBufferStreamTest (DERBY-3796).
Patch contributed by Erlend Birkenes
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@682531 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3818-f6967015 | DERBY-3818 (partial): Whitespace and formatting changes only.
Patch file: n/a
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@707103 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3819-05ef8bdc | DERBY-3819 ; 'Expected Table Scan ResultSet for T3' in test_predicatePushdown(....PredicatePushdownTest)
Causing a number of asserts to be skipped when using a 64 bit platform.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@807733 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3823-7b429a31 | DERBY-3823 NullPointerException in stress.multi test
Adding a test case showing that in case of a network server, an open resulset's metadata can get changed underneath it but it is
not reflected in the metadata. The test creates a table with one of column as varchar(5). It inserts 1000 rows and then
opens a reulset on that table with varchar column as one of the columns. The test verifies that the reulset's metadata
at this point shows the length of the column as 5. Next, while the resulset is still open, the tests does an ALTER TABLE
to increase the varchar column's length to 8. In case of embedded mode, this fails because of the open resulset. In case
of network server, because of prefetching of rows, the ALTER TABLE is allowed but when the test gets the resulset's
metadata again and checks the length of varchar column, it still shows the length to be 5 rather than 8.
There are couple other jiras related to network server prefetching, namely, DERBY-3839 and DERBY-4373.
Once DERBY-3823 is fixed, we should see the change in metadata reflected in resultset's metadata. A fix for DERBY-3823 will
cause the following the test added here to fail. Right now, the new test accepts the incorrect metadata length obtained
through the resultset's metadata after ALTER TABLE has been performed in network server mode.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@1182570 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3823-e8fb6d60 | DERBY-3823 NullPointerException in stress.multi test
This patch fixes a race condition in
EmbedPreparedStatement#getMetaData: if we are trying to retrieve the
metadata for a prepared statement while it is being recompiled there
is a time window during which the activation class is null. The
existing code could therefore cause an NPE. The new code plugs the
race condition. This NPE led to intermittent errors in
stressTestMulti.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@1183192 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/jdbc/EmbedPreparedStatement.java",
"hunks": [
{
"added": [
"import org.apache.derby.iapi.services.loader.GeneratedClass;"
],
"header": "@@ -64,6 +64,7 @@ import java.sql.Types;",
"removed": []
},
{
... |
derby-DERBY-3825-925b07bc | DERBY-3825: StoreStreamClob.getReader(charPos) performs poorly.
Made the repositioning logic easier to read.
Patch file: derby-3825-3a-simplification.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@710033 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/jdbc/UTF8Reader.java",
"hunks": [
{
"added": [
" if (requestedCharPos <= readerCharCount - charactersInBuffer) {",
" // The stream must be reset, because the requested position is",
" // bef... |
derby-DERBY-3825-ab2037f8 | DERBY-3825: StoreStreamClob.getReader(charPos) performs poorly.
Added repositioning functionality to UTF8Reader - ordered after increasing cost:
a) Reposition within current character buffer (small hops forwards and potentially backwards - in range 1 char to 8K chars)
b) Forward stream from current position (hops forwards)
c) Reset stream and skip data (hops backwards)
In addition the new method getInternalReader was added to InternalClob. It takes advantage of the repositioning capability of UTF8Reader.
Clob.getSubString uses the new functionality when the Clob is represented by a StoreStreamClob.
Previously only mechanism c) was used.
Patch file: derby-3825-2b-internalReader_repositioning.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@704547 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/jdbc/EmbedClob.java",
"hunks": [
{
"added": [
" Reader reader;",
" try {",
" reader = this.clob.getInternalReader(pos);",
" } catch (EOFException eofe) {",
... |
derby-DERBY-3831-a90874a5 | DERBY-3831; junit.RuntimeStatisticsParser doesn't fully distinguish names of table or index.
patch contributed by Junjie Peng
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@685733 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/testing/org/apache/derbyTesting/junit/RuntimeStatisticsParser.java",
"hunks": [
{
"added": [
" tableName + \" \")!= -1);"
],
"header": "@@ -168,7 +168,7 @@ public class RuntimeStatisticsParser {",
"removed": [
" ... |
derby-DERBY-3840-f23a20f8 | DERBY-3840 The test code executes java processes by just executing java instead of using a full path. This may cause the wrong java to be picked up
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@693522 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/testing/org/apache/derbyTesting/junit/NetworkServerTestSetup.java",
"hunks": [
{
"added": [
" al.add( BaseTestCase.getJavaExecutableName() );"
],
"header": "@@ -249,7 +249,7 @@ final public class NetworkServerTestSetup extends BaseTestSetup {",
... |
derby-DERBY-3844-b61d6349 | DERBY-3844: ASSERT failure in BasePage.unlatch() when running LobStreamsTest
Disallows calling getBlob or getClob more than once on a given LOB column
(on the same result set row). See release note for details, this change may
break existing applications.
Patch file: derby-3844-1c-disallow_multiple_accesses.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@911793 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/client/org/apache/derby/client/am/ResultSet.java",
"hunks": [
{
"added": [
"import java.util.Arrays;"
],
"header": "@@ -25,9 +25,9 @@ import java.io.IOException;",
"removed": [
"import org.apache.derby.shared.common.i18n.MessageUtil;"... |
derby-DERBY-3844-c1832a32 | DERBY-3844: ASSERT failure in BasePage.unlatch() when running LobStreamsTest
Rewrote test to only call getX once per column per row, as calling for instance
getBlob twice on a given column on the same row will be disallowed.
Also removed some unused code.
Patch file: derby-3844-2a-jdbcdriver_test_rewrite.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@910481 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3845-d8644b2a | DERBY-3845: Problems running org.apache.derbyTesting.system.optimizer.RunOptimizerTest
Make sure the database is initialized also when -mode is specified on
the command line.
Patch contributed by Ole Solberg <Ole.Solberg@Sun.COM>.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@691583 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3850-8106edc9 | DERBY-3850: Remove unneeded workarounds for DERBY-177 and DERBY-3693
Removed the wait parameter from methods called from
SPSDescriptor.updateSYSSTATEMENTS() since waiting is prevented by
another mechanism now.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@692495 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/sql/dictionary/DataDictionary.java",
"hunks": [
{
"added": [
"\t\tTransactionController\ttc"
],
"header": "@@ -1072,15 +1072,13 @@ public interface DataDictionary",
"removed": [
"\t * @param wait\t\t\tTo w... |
derby-DERBY-3850-a9215529 | DERBY-3850: Remove unneeded workarounds for DERBY-177 and DERBY-3693
Removed the wait parameter from TabInfoImpl.updateRow(). The method
only had two callers, both of which called it with
wait=true. updateRow() passed the parameter on to openForUpdate() in
RowChanger, but that method is sometimes called with wait=false, so
the parameter couldn't be removed from that method.
Also removed an unused variable and some unused imports.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@695244 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/catalog/TabInfoImpl.java",
"hunks": [
{
"added": [],
"header": "@@ -22,18 +22,14 @@",
"removed": [
"import org.apache.derby.iapi.services.context.ContextService;",
"import org.apache.derby.iapi.sql.conn.Langua... |
derby-DERBY-3853-d1661b20 | DERBY-3853: Behaviour of setTypeMap() differs between embedded and client
Changed the client driver to match the embedded driver.
Patch contributed by Yun Lee <yun.lee.bj@gmail.com>.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@764217 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3856-79fec783 | DERBY-3856: difference between Embedded vs DerbyNetClient in format of
return from timestamp(cast(? as varchar(32)))
Stop caching the original input string in the parse methods of SQLDate
and SQLTimestamp, and instead generate (and cache) a normalized
datetime string on the first call to getString().
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@952581 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/types/SQLDate.java",
"hunks": [
{
"added": [],
"header": "@@ -471,7 +471,6 @@ public final class SQLDate extends DataType",
"removed": [
" valueString = parser.getTrimmedString();"
]
},
{
... |
derby-DERBY-3863-40a45842 | DERBY-3863; improve test for import export using ij
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@691506 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-387-5bd651f5 | DERBY-387: Fix Simple Network Client sample to work correctly by fixing
database name. Contributed by Rajesh Kartha
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@375715 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/demo/nserverdemo/SimpleNetworkClientSample.java",
"hunks": [
{
"added": [
"import java.lang.reflect.InvocationTargetException;",
"import java.lang.reflect.Method;",
"import java.sql.Connection;",
"import java.sql.DriverManager;",
... |
derby-DERBY-3870-044afae1 | DERBY-3870: Concurrent Inserts of rows with XML data results in an exception
Store XML utility instance in the activation instead of in the
compiled plan, so that it's never accessed concurrently by multiple
threads.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@1125305 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/types/SqlXmlUtil.java",
"hunks": [
{
"added": [],
"header": "@@ -23,8 +23,6 @@ package org.apache.derby.iapi.types;",
"removed": [
"import org.apache.derby.iapi.services.io.Formatable;",
"import org.apache.derby.i... |
derby-DERBY-3870-0f64b702 | DERBY-5289 Unable to boot 10.5.1.1 database - fails during soft/hard upgrade process for a new version number while trying to drop jdbc metadata
Checking testcase for DERBY-5289. In trunk theDERBY-3870 fix contributed by Knut Anders Hatlen fixed the issue so no code change is needed. Just the portion of DERBY-3870 that is relevant to DERBY-5289 will be backported to the other branches.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@1139449 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3870-118ac264 | DERBY-3870: Concurrent Inserts of rows with XML data results in an exception
Added a test case for the bug (disabled for now).
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@1101839 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3870-2a09eb57 | DERBY-3870: Concurrent Inserts of rows with XML data results in an exception
Remove lazy initialization of the field that holds the cached
SqlXmlUtil instance. This simplifies the generated byte code. It also
removes the need for an explicit syntax check of the XML query during
the bind phase, as the earlier initialization ensures that syntax
errors will be detected at compile time.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@1126358 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/compile/BinaryOperatorNode.java",
"hunks": [
{
"added": [],
"header": "@@ -31,7 +31,6 @@ import org.apache.derby.iapi.services.compiler.LocalField;",
"removed": [
"import org.apache.derby.iapi.types.SqlXmlUtil;"
... |
derby-DERBY-3870-5fc727ca | DERBY-3870: Concurrent Inserts of rows with XML data results in an exception
Allow databases that contain triggers with XML operators to be
upgraded from a version without the fix for this issue to a fixed
version. It used to fail because it couldn't deserialize the old plan
for the trigger. Fix it by not trying to deserialize the plan before
invalidating it.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@1132546 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/catalog/DataDictionaryImpl.java",
"hunks": [
{
"added": [
" * The returned descriptors don't contain the compiled statement, so it",
" * it safe to call this method during upgrade when it isn't known if the",
... |
derby-DERBY-3870-d09782a3 | DERBY-3870: Concurrent Inserts of rows with XML data results in an exception
Remove the unneeded indirection via SqlXmlExecutor. This reduces the
amount of code and removes one object allocation per row when using an
XML operator.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@1127883 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/types/XML.java",
"hunks": [
{
"added": [
" * @param stringValue The string value to check."
],
"header": "@@ -598,7 +598,7 @@ public class XML",
"removed": [
" * @param text The string value to che... |
derby-DERBY-3870-ed7f8b90 | DERBY-6634: Improve test coverage of SqlXmlUtil.java
Remove dead code from the time when SqlXmlUtil implemented the
Formatable interface (before DERBY-3870).
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@1605285 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/types/SqlXmlUtil.java",
"hunks": [
{
"added": [],
"header": "@@ -128,12 +128,6 @@ public class SqlXmlUtil",
"removed": [
" // Used to recompile the XPath expression when this formatable",
" // object is reco... |
derby-DERBY-3871-0c5c5aa1 | DERBY-3871: EmbedBlob.setBytes returns incorrect insertion count.
Made EmbedBlob.setBytes return the number of bytes inserted, instead of returning the Blob position after the insert.
Improved some JavaDoc comments.
Added regression tests.
Patch file: derby-3871-1a_insertion_count.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@701372 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/jdbc/EmbedBlob.java",
"hunks": [
{
"added": [
" * Writes the given array of bytes to the BLOB value that this Blob object",
" * represents, starting at position pos, and returns the number of bytes",
" * wri... |
derby-DERBY-3872-f36770cc | DERBY-3872
The NPE in this jira entry was caused by the missing overwrite of accept() method in IndexToBaseRowNode.
Because of the missing code, the additional layer of VirtualColumn node over ResultColumn was not
happening for the where clause in HAVING. Once the accept method was added to IndexToBaseRowNode, the
VirtualColumn on top of the ResultColumn got the correct resultset number associated with it and at the
code generation time, we start referencing the correct resultset rather than the one associated with
the JOIN clause. Thanks a ton to Army and Bryan on this jira entry for their help.
I have added a test case for this in lang/GroupByTest.java
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@705037 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/compile/IndexToBaseRowNode.java",
"hunks": [
{
"added": [
"import org.apache.derby.iapi.sql.compile.Visitable;",
"import org.apache.derby.iapi.sql.compile.Visitor;"
],
"header": "@@ -27,6 +27,8 @@ import org.a... |
derby-DERBY-3875-c33b0cfc | DERBY-3875; closing containers after a problem has been found allows restoreFrom to work correctly.
Patch contributions by Jason McLaurin, Kristian Waagan and Myrna van Lunteren.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@703246 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/testing/org/apache/derbyTesting/junit/Utilities.java",
"hunks": [
{
"added": [
"import java.util.StringTokenizer;"
],
"header": "@@ -33,6 +33,7 @@ import java.security.PrivilegedExceptionAction;",
"removed": []
},
{
"added":... |
derby-DERBY-3878-42526fd1 | DERBY-3878: Replication: stopSlave does not close serversocket when master has crashed
Use try/finally to ensure that the sockets are closed when closing the
socket streams fails.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@707591 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/store/replication/net/SocketConnection.java",
"hunks": [
{
"added": [
" // If the other party has crashed, closing the streams may fail (at",
" // least the output stream since its close() method calls flush()).",
... |
derby-DERBY-388-851bcbb2 | Port fix for DERBY-388 into 10.1 branch. Address intermittent failures when executing trigger statements caused by references to internal SQL formats.
Fix originally submitted to 10.0 branch by Army Brown (qozinx@sbcglobal.net)
git-svn-id: https://svn.apache.org/repos/asf/incubator/derby/code/trunk@219115 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/GenericPreparedStatement.java",
"hunks": [
{
"added": [
"\t\t\tif (!spsAction) {",
"\t\t\t// only re-prepare if this isn't an SPS for a trigger-action;",
"\t\t\t// if it _is_ an SPS for a trigger action, then we can... |
derby-DERBY-388-8f235488 | DERBY-388: Add a test case to cover this patch submitted earlier.
Submitted by Army Brown (qozinx@sbcglobal.net)
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@226896 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3880-2e7e8f6d | DERBY-4698 Simple query with HAVING clause crashes with NullPointerException
Patch derby-4698-2. The case of column references in HAVING clauses
being wrong after JOIN flattening was initially solved by
DERBY-3880. That solution was partial in that it can sometimes happen
too late.
This patch changes the fix-up of column references in a having clause
after join flattening to the same point in time as that of other
column references that need fix-up after the flattening (rcl, column
references in join predicates and group by clauses). Thus the fixup is
moved from the modifyaccesspath phase to the preprocess phase.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@956234 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/compile/FromList.java",
"hunks": [
{
"added": [
" * @param havingClause The HAVING clause, if any",
" GroupByList gbl,",
" ValueNode havingC... |
derby-DERBY-3880-8d13a1f3 | DERBY-3880 NPE on a query with having clause involving a join
remap expression for AggregateNode operand if the JoinNode has been flattened.
Fix contributed by Army Brown
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@711321 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3882-4880a417 | DERBY-3882: Expensive cursor name lookup in network server
Statements executed via the network client are always given cursor
names on the server, even if the user has not set a cursor name. When
a statement has a cursor name, the embedded driver will check on each
execution that there is no other statement in the same connection with
the same cursor name and an open result set. To perform this check,
the list of activations in the connection is traversed and each cursor
name is compared with the cursor name of the statement to be executed.
If the number of open statements in the connection is high, which is
very likely if ClientConnectionPoolDataSource with the JDBC statement
cache is used, traversing the list of activations and performing
string comparisons may become expensive.
This patch attempts to make this operation cheaper without performing
a full rewrite and use a new data structure. It exploits that the most
common implementations of java.lang.String cache the hash code, so
calling hashCode() on the cursor names in the list will simply read
the value from an int field after warm-up. By checking if the hash
codes of the cursor names match first, we can avoid many of the string
comparisons because we know that strings with different hash codes
cannot be equal.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@818807 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/conn/GenericLanguageConnectionContext.java",
"hunks": [
{
"added": [
" int cursorHash = cursorName.hashCode();",
""
],
"header": "@@ -876,6 +876,8 @@ public class GenericLanguageConnectionContext",
... |
derby-DERBY-3884-ebad814e | DERBY-3877 SQL roles: build support for dblook
Patch derby-3877-2, which adds basic support for roles in dblook (but
see DERBY-3884), adds test cases, and also enables the old test
harness to shutdown the server using credentials when required due to
authentication being enabled (needed by modified test for dblook).
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@700295 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/testing/org/apache/derbyTesting/functionTests/harness/NetServer.java",
"hunks": [
{
"added": [
"\tString appsRequiredPassword;"
],
"header": "@@ -43,6 +43,7 @@ public class NetServer",
"removed": []
},
{
"added": [
... |
derby-DERBY-3886-7cbf0216 | DERBY-3886 SQL roles: ij show enabled and settable roles
Adds a VTI table function SYSCS_DIAG.ENABLED_ROLES and two new ij
commands; "show settable_roles" and "show enabled_roles", plus new
tests. Also tweaks ScriptTestCase to be able to run scripts that need
SQL authorization.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@702266 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3887-7f4445db | DERBY-3887 Embedded Derby fails under JBoss because of JMX-related conflicts
Backed out JMX related changes for DERBY-3745.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@784831 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/services/jmx/JMXManagementService.java",
"hunks": [
{
"added": [],
"header": "@@ -163,43 +163,7 @@ public final class JMXManagementService implements ManagementService, ModuleCont",
"removed": [
" //DERBY-3745 We wan... |
derby-DERBY-3888-b5392dbe | DERBY-3888: ALTER TABLE ... ADD COLUMN cannot add identity columns
Enable support for adding identity columns with ALTER TABLE. It is
only enabled if identity columns are backed by sequences. That is, the
database format has to be 10.11 or higher.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@1626141 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/execute/AlterTableConstantAction.java",
"hunks": [
{
"added": [],
"header": "@@ -24,7 +24,6 @@ package org.apache.derby.impl.sql.execute;",
"removed": [
"import org.apache.derby.catalog.DefaultInfo;"
]
},
... |
derby-DERBY-3889-2fb5c8dd | DERBY-3889: LOBStreamControl.truncate() doesn't delete temporary files
Created helper method for closing and deleting temporary files to make
it less likely that some of the required operations are
forgotten. Also moved the lobFile field into the LOBFile class to
reduce the risk of getting inconsistencies between the fields tmpFile
and lobFile.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@704010 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/jdbc/LOBFile.java",
"hunks": [
{
"added": [
" /** The temporary file where the contents of the LOB should be stored. */",
" private final StorageFile storageFile;",
"",
" /** An object giving random a... |
derby-DERBY-389-c62aaa25 | DERBY-389
Fix a hang accessing the statement cache in network server stress test. The change was to remove the synchronization on the statement cache from the removeStatement method of GenericLanguageConnectionContext. It is not needed because the statement cache handles its own synchronization. Also changes the javadoc to remove some incorrect info about temp tables.
git-svn-id: https://svn.apache.org/repos/asf/incubator/derby/code/trunk@201792 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/conn/GenericLanguageConnectionContext.java",
"hunks": [
{
"added": [
"\t* This method will remove a statement from the statement cache.",
"\t* It will be called, for example, if there is an exception preparing",
... |
derby-DERBY-389-df5e1dd0 | DERBY-389 Fix nullSQLText.java for J2ME
Attaching a patch for the test jdbcapi/nullSQLText.java. This test fails in J2ME because it uses a stored procedure with server-side JDBC. As I understand, the purpose of the test is only to test any stored procedure call. So I replaced the stored procedure in this test with a procedure in org.apache.derbyTesting.functionTests.util.ProcedureTest. Also changed the master files.
With this patch, I have run jdbcapi/nullSQLText.java in embedded, client and jcc frameworks. Also run this test in CDC/FP. This patch changes only tests. Please review/commit this patch.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@330705 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3890-9cc8ad0a | DERBY-3890: Replication: NPE for startSlave of encrypted database
Removes NPE for replication of encrypted databases by setting
RawStoreFactory in LogFactory before calling SlaveFactory#startSlave.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@708510 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/store/raw/RawStore.java",
"hunks": [
{
"added": [
" // RawStoreFactory is used by LogFactory.recover() and by",
" // SlaveFactory.startSlave (for the SlaveFactory case, it is",
" // only used if the... |
derby-DERBY-3897-e3883f5f | DERBY-3897 SQLSessionContext not correctly initialized in some non-method call nested contexts
Patch derby-3897-3, which sets up the SQL session context correctly
also for substatements. See javadoc for
LanguageConnectionContext#setupSubStatementSessionContext for an
enumeration of these cases. Also adds test cases for this.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@703295 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/sql/PreparedStatement.java",
"hunks": [
{
"added": [
"\t * Execute the PreparedStatement and return results, used for top level",
"\t * statements (not substatements) in a connection."
],
"header": "@@ -93,7 +93,8... |
derby-DERBY-3898-24400cd2 | DERBY-3898: Blob.setBytes differs between embedded and client driver
when the specified length is invalid
Added fix and test case for a remaining corner case: When the sum of
offset and length is greater than Integer.MAX_VALUE, the client driver
silently ignores the error whereas the embedded driver fails with an
IndexOutOfBoundsException.
The unexpected results are caused by a check for
offset + len > bytes.length
where offset+len overflows and evaluates to a negative value. The fix
changes this condition to the equivalent
len > bytes.length - offset
which won't overflow (because both bytes.length and offset are known
to be non-negative at this point in the code, and subtracting one
non-negative int from another is guaranteed to result in a value in
the range [-Integer.MAX_VALUE, Integer.MAX_VALUE]).
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@986345 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-390-c47d471f | DERBY-390: Patch to handle case-sensitive SQL identifiers correctly.
Import/export procedure parameters for table names, schema names, columns
names should be passed in the case-sensitive form if they are quoted identfiers
and in upper case if they are not quoted SQL identifiers. Import/export will
generate insert/select statements with quoted table names, schema name and
column names to be execute on the database after this patch.
committed on behalf of: Suresh Thalamati
git-svn-id: https://svn.apache.org/repos/asf/incubator/derby/code/trunk@208770 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/load/ColumnInfo.java",
"hunks": [
{
"added": [
"",
"\t\tthis.schemaName = sName;",
"\t\tthis.tableName = tName;",
"\t\t\t//eg: C2 , C1 , C3"
],
"header": "@@ -80,13 +80,14 @@ class ColumnInfo ... |
derby-DERBY-3902-47509e86 | DERBY-3902; adjust orphaned message strings;
- move J107, J108, J109 to common MessageId.java
- remove references to 08000.S.1 and XJ102 from message.xml & mes*.properties
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@704290 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/client/org/apache/derby/client/am/SqlException.java",
"hunks": [
{
"added": [
"import org.apache.derby.shared.common.reference.MessageId;"
],
"header": "@@ -26,6 +26,7 @@ import java.util.TreeMap;",
"removed": []
},
{
"added... |
derby-DERBY-3902-648e3486 | DERBY-3902; correct messages.xml and translated files for SQLState 2003.S.4
(was accidentally listed as 2004.S.4 - see also comment in DERBY-1567).
Also fix up MessageBundleTest to ignore two sqlstates that are not exposed to
the user so don't need text.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@703259 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3904-57191b46 | DERBY-3904: NPE on left join with aggregate
The issue involves a very special optimization that is performed
for MIN and MAX queries in which we may be able to use an index
to go directly to the lowest/highest value of the desired column.
For example, in the query
SELECT MAX(d1) FROM t1
if there is an index on d1, we can use that index to retrieve
the max value very rapidly. In order to incorporate this
optimization, the following conditions must be met:
- No group by
- One of:
- min/max(ColumnReference) is only aggregate && source is
ordered on the ColumnReference
- min/max(ConstantNode)
The optimization of the other way around (min with desc index or
max with asc index) has the same restrictions with the additional
temporary restriction of no qualifications at all (because
we don't have true backward scans).
The source of the data must also be "simple" (not a result of a join),
and the NullPointerException occurred during the code that tried
to establish the above conditions because it wasn't thoroughly
enough excluding the join case. In the query:
SELECT MAX( T1.D1 ) AS D FROM T1 LEFT JOIN T2 ON T1.D1 = T2.D2
WHERE T2.D2 IS NULL
the code in GroupByNode.considerPostOptimizeOptimizations was
trying to traverse the AccessPathImpl to find the index scan
information, but for this LEFT JOIN case there is an AccessPathImpl
but no index scan information, because there is a join being
performed, not an index scan.
The solution is to examine the AccessPathImpl more carefully,
and only search the index scan information if an index scan
is actually present.
Also added a few tests, including an enhancement to the test
library's RuntimeStatisticsParser so that it can determine
if a Last Key Index Scan is being performed by the query.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@708002 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/testing/org/apache/derbyTesting/junit/RuntimeStatisticsParser.java",
"hunks": [
{
"added": [
"\tprivate final boolean lastKeyIndexScan;"
],
"header": "@@ -31,6 +31,7 @@ public class RuntimeStatisticsParser {",
"removed": []
},
{
... |
derby-DERBY-3905-d7731394 | DERBY-3905 Failed tests should save the database off to the fail directory
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@704964 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3907-51c45b8b | DERBY-3907: Save useful length information for Clobs in store.
Cleanup of ReaderToUTF8Stream, which has to deal with the header in the streams being passed in to Derby. Changes:
o Simplified constructors.
o Added JavaDoc and comments.
o Removed unused imports.
o Removed instance variable maximumLength.
o Added more information to the error messages for truncation.
o Added CHAR as a truncatable string data type.
o Removed "throws IOException" from close.
Updated the test to pass inn a valid type name to the constructor.
Patch file: derby-3907-3b-readertoutf8stream_cleanup.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@732676 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/types/ReaderToUTF8Stream.java",
"hunks": [
{
"added": [
"import org.apache.derby.iapi.services.sanity.SanityManager;",
" * Converts the characters served by a {@code java.io.Reader} to a stream",
" * returning the data ... |
derby-DERBY-3907-52625a50 | DERBY-3907: Save useful length information for Clobs in store.
Enabled the new header format for Clobs.
Description:
o ClobStreamHeaderGenerator
Enabled the callback mechanism to inform the DVD about whether the database
is accessed in soft upgrade mode or not.
o SQLChar
Added method writeClobUTF, which writes a Clob to the on-disk format.
Added method readExternalClobFromArray.
o SQLClob
Added a variable to tell if the database is accessed in soft upgrade mode
or not. It is used to reduce object creation (header generators) and to
avoid consulting the data dictionary as much. It requires that the DVDs are
reused, and I'm sure this can be further optimized.
Implemented getLength, which returns the length of the Clob in one of
three ways:
1) Delegate to SQLChar if the Clob has been materialized.
2) Read length from stream header if present.
3) Decode the whole stream.
Updated getStreamWithDescriptor to deal with both header formats.
Made restoreToNull nullify the character stream descriptor.
Implemented writeExternal.
Implemented getStreamHeaderGenerator, which will return one of two shared
generator instances it is known if the database is accessed in soft upgrade
mode or not. If unknown, a new generator instance is created, which will
determine the mode when the header is asked for.
Implemented investigateHeader, which decodes a stream header.
Implemented readExternal.
Implemented readExternalFromArray.
Implemented utility method rewindStream, which resets the stream and skips
the number of characters specified.
Added a utility class for holding header information (currently only length).
o StreamHeaderHolder
Deleted the class, it is no longer used.
NOTE: Databases created with this revision (or later) containing Clobs, cannot
be accessed by earlier trunk revisions.
Patch file: derby-3907-7a3-use_new_header_format.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@738408 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/types/SQLChar.java",
"hunks": [
{
"added": [
" /**",
" * Writes the header and the user data for a CLOB to the destination stream.",
" *",
" * @param out destination stream",
" * @... |
derby-DERBY-3907-6f4c92af | DERBY-3907 (partial): Save useful length information for Clobs in store.
Started using the new framework for handling stream headers for string data
values. The behavior regarding stream headers is kept unchanged, but the code
is now ready to deal with multiple stream header formats.
Short description:
* EmbedResultSet & EmbedPreparesStatement
Adjusted code to use the new interface method and pass in the correct
class to the ReaderToUTF8Stream constructor.
Note the special case of telling the DVD/generator if the database being
accessed is in soft upgrade mode in EmbedResultSet.
* ArrayInputStream
The stream header is no longer read inside readDerbyUTF.
* ReaderToUTF8Stream
Adjusted code to use the new StreamHeaderGenerator interface, and made the
stream count the number of characters encountered.
If possible, the header is updated when the stream has been drained.
* StringDataValue
Added methods getStreamHeaderGenerator and setSoftUpgradeMode.
* SQLChar
Refactoring in preparation for handling multiple stream header formats.
Pulled common code out into writeUTF. The header generator is now
repsonsible for writing both the header bytes and an EOF marker if required.
Made a second readExternal method, which is not reading the stream header
format. This must now be done outside of this method and any length
information is passed in as arguments.
Implemented the new methods in StringDataValue.
* SQLClob
Adjusted a single call to ReaderToUTF8Stream.
* UTF8UtilTest
Adjusted code invoking the new ReaderToUTF8Stream constructor.
Patch file: derby-3907-7a2-use_new_framework.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@736636 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/services/io/ArrayInputStream.java",
"hunks": [
{
"added": [],
"header": "@@ -23,7 +23,6 @@ package org.apache.derby.iapi.services.io;",
"removed": [
"import java.io.ObjectInput;"
]
},
{
"added": ... |
derby-DERBY-3907-7672693c | DERBY-3907: Save useful length information for Clobs in store.
These changes make Derby start using the stream descriptor object to
correctly handle streams from store. This is done to be able to use
several header formats, where a varying number of bytes are used.
Description of changes:
o EmbedClob
Changed constructor to take a StringDataValue instead of a
DataValueDescriptor. Updated call to the StoreStreamClob constructor.
o EmbedResultSet
Started using the getStreamWithDescriptor method and updated invocations of
the UTF8Reader constructor.
o StoreStreamClob
Added a CharacterStreamDescriptor (CSD), and made the constructor take one
as an argument. Adapted the class to use a CSD.
o UTF8Reader
Updated some comments.
Fixed bug where the header length wasn't added to the byte length of the
stream, and updated the class appropriately (adjusted utfCount, fixed the
reset routine). Made sure the header bytes are skipped (either by skipping
them in the constructor or by adjusting the position to on the next
reposition).
o ResultSetStreamTest
Added a test for maxFieldSize, where truncation have to happen.
o Various tests
Adjusted tests to run with the new implementation.
Patch file: derby-3907-5a-use_getStreamWithDescriptor.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@734630 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/jdbc/EmbedClob.java",
"hunks": [
{
"added": [
"import org.apache.derby.iapi.jdbc.CharacterStreamDescriptor;",
"import org.apache.derby.iapi.types.StringDataValue;"
],
"header": "@@ -25,15 +25,11 @@ package org.apa... |
derby-DERBY-3907-7af67265 | DERBY-3907 (partial): Save useful length information for Clobs in store.
Added the framework required to handle multiple stream header formats for
stream sources. Note that handling of the new header format is not yet added,
so the code should behave as before, using only the old header format.
A description of the changes:
o EmbedResultSet and EmbedPreparedStatement
Started using the new ReaderToUTF8Stream constructor, where the stream
header is passed in. Also started to treat the DataValueDescriptor as a
StringDataValue, which should always be the case at this point in the code.
o ReaderToUTF8Stream
Added field 'header', which holds a StreamHeaderHolder coming from a
StringDataValue object. Updated the constructors with a new argument. The
first execution of fillBuffer now uses the header holder to obtain the
header length, and the header holder object is consulted when checking if
the header can be updated with the length after the application stream has
been drained. Note that updating the header with a character count is not
yet supported.
o StringDataValue
Added new method generateStreamHeader.
o SQLChar
Implemented generateStreamHeader, which always return a header for a stream
with unknown length (see the constant).
o SQLClob
Added a constant for a 10.5 stream header holder representing a stream with
unknown character length. Also updated the use of the ReaderToUTF8Stream
constructor.
o StreamHeaderHolder (new file)
Holder object for a stream header, containing the header itself and the
following additional information; "instructions" on how to update the header
with a new length, if the length is expected to be in number of bytes or
characters, and if an EOF marker is expected to be appended to the stream.
o UTF8UTilTest
Updated usage of the ReaderToUTF8Stream constructor, and replaced the
hardcoded byte count to skip with a call to the header holder object.
o jdbc4.ClobTest
Added some simple tests inserting and fetching Clobs to test the basics of
stream header handling.
o StreamTruncationTest (new file)
New test testing truncation of string data values when they are inserted as
streams.
Patch file: derby-3907-2c-header_write_preparation.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@734065 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/types/ReaderToUTF8Stream.java",
"hunks": [
{
"added": [
" /**",
" * The stream header to use for this stream.",
" * <p>",
" * The holder object is immutable, and the header should not have to be... |
derby-DERBY-3907-cf0fdc4f | DERBY-3907 (partial): Save useful length information for Clobs in store.
Added StringDataValue.getStreamWithDescriptor().
It it intended to be used when getting a stream from a StringDataValue to be
used with a Clob object, or with streaming of string data values in general.
The DVD is responsible for returning a correct descriptor for the raw stream.
The descriptor is in turn used by other classes to correctly configure
themselves with respect to data offsets, buffering, repositioning and so on.
Changes:
o CharacterStreamDescriptor
Added a toString method and more verbose assert-messages.
o StringDataValue
Added method 'CharacterStreamDescriptor getStreamWithDescriptor()'.
o SQLChar
Made setStream non-final so it can be overridden in SQLClob.
Added default implementation of getStreamWithDescriptor that always returns
null. This means that all non-Clob string data types will be handled as
strings instead of streams in situations where a stream is requested through
getStreamWithDescriptor. This might be changed.
Made throwStreamingIOException protected to access it from SQLClob.
o SQLClob
Implemented getStreamWithDescriptor, handling the old 2-byte format only.
Overrid setStream to discard the stream descriptor when a new stream is set
for the DVD.
Patch file: derby-3907-4a-add_getStreamWithDescriptor.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@734148 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/jdbc/CharacterStreamDescriptor.java",
"hunks": [
{
"added": [
" SanityManager.ASSERT(curBytePos >= 0, \"Negative curBytePos\");",
" curCharPos == BEFORE_FIRST, \"Invalid curCharPos \" +",
... |
derby-DERBY-3907-dfdebd5e | DERBY-3907 (partial): Save useful length information for Clobs in store.
Added an interface for a stream header generator object, and two implementations; one for Clob and one for the non-Clob string types (CHAR, VARCHAR and LONG VARCHAR).
To support pre 10.5 databases in soft upgrade mode, the Clob stream header generator can generate both the new header format and the old one.
Note that the Clob header generator depends on either knowing up front if it is being used in soft upgrade mode, or to be able to determine this through a database context.
Patch file: derby-3907-7a1-write_new_header_format.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@736612 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/types/StreamHeaderGenerator.java",
"hunks": [
{
"added": [
"/*",
"",
" Derby - Class org.apache.derby.iapi.types.StreamHeaderGenerator",
"",
" Licensed to the Apache Software Foundation (ASF) und... |
derby-DERBY-3909-db8b20bd | DERBY-3909: Race condition in NetXAResource.removeXaresFromSameRMchain()
Removed the code that had race conditions since the data structures it
touched weren't actually used for anything.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@704904 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/client/org/apache/derby/client/net/NetConnection.java",
"hunks": [
{
"added": [],
"header": "@@ -1753,9 +1753,6 @@ public class NetConnection extends org.apache.derby.client.am.Connection {",
"removed": [
" if (xares_ != null) {",
" ... |
derby-DERBY-3917-52d76403 | DERBY-3917; skip fixture testCurrentRoleInWeirdContexts and 1 test case in
fixture testDefaultCurrentRole in lang.RolesConferredPrivilegesTest with
JSR169.
Patch contributed by Dag H. Wanvik.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@713533 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3922-d4f93087 | DERBY-3922: Support for adding generated columns via ALTER TABLE.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@709152 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/compile/UpdateNode.java",
"hunks": [
{
"added": [
"import java.util.HashSet;"
],
"header": "@@ -81,6 +81,7 @@ import java.lang.reflect.Modifier;",
"removed": []
},
{
"added": [
" ... |
derby-DERBY-3925-fc061da9 | DERBY-3925 - testMetaDataQueryRunInSYScompilationSchema(.....upgradeTests.Changes10_4) fails on CVM/phoneME
Use 'territory=en' instead of 'territory=no' if the 'no' locale is not available.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@823555 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3926-200a5ea7 | DERBY-3926
Following is the patch description for DERBY-3926.
The problem with the trunk codeline is that when optimizer goes through optimizables in a join order, it only looks at those optimizables individually to decide whether sorting can be avoided on them or not. That approach leaves out few queries which require sorting but do not get sorted. The decision for avoiding sorting should also include relationship between the optimizables in a given join order. Following query demonstrates the trunk problem
SELECT table1.id, table2.value, table3.value FROM --DERBY-PROPERTIES joinOrder=FIXED
table3 -- DERBY-PROPERTIES index=nonUniqueOnValue_Table3
, table2 -- DERBY-PROPERTIES index=nonUniqueOnValue_Table2
, table1
WHERE table1.id=table2.id AND table2.name='PageSequenceId'
AND table1.id=table3.id
AND table3.name='PostComponentId'
AND table3.value='21857' ORDER BY table2.value;
In the query above, when optimizer is considering [table3, table2, -1] join order, it determines that sorting can be avoided on this join order because the order by column table2.value is already covered by the index nonUniqueOnValue_Table2. It does not see that the outermost optimizable table3 will qualify more than one row and hence it will be a multi-row resulset and for each one of those rows, we will be doing a scan into table2. In other words, there will be multiple scans into table2(and the rows returned by each one of those scans will be ordered on table2.value) but the collective rows from those multiple scans are not necessarily going to be ordered on table2.value. This patch is attempting to fix that problem.
Currently, in trunk, a column is marked always ordered during a query processing when the optimizer finds that there is constant comparison predicate on the order by column. If the column does not have a constant predicate (as in our example above), we next see if we are using an index which will provide the required ordering on column (which is true in our case. The required ordering on table2.value is provided by the index nonUniqueOnValue_Table2). But as we can see in the query above, this index coverage is not enough to say that sorting is not needed. We need to add 2 more conditions before we can decide to avoid the sorting. One of those cases is 1)if the order by column does not belong to the outermost optimizable, then check if the order by column's optimizable is a one-row resultset. If yes, then it will be safe for the optimizer to avoid the sorting. The second case to consider is 2)if the order by column does not belong to the outermost optimizable, then check if the order by column's optimizable is multi-row resultset BUT all the outer optimizables are one-row resulsets. If either of these 2 additional conditions are satisfied then optimizer can choose to avoid the sorting. Otherwise
sorting should be added to the query plan. The example query above does not satisfy the 2 additional checks and hence sorting should be done as part of the query plan.
The changes for the 1)check above has gone into OrderbyByList.sortRequired(RowOrdering, JBitSet, OptimizableList). The implementation of this change just required us to check the outer optimizables to be one row since the order by column's optimizable is not one row. If outer optimizables are all one-row, then we say that sorting can be avoided. Otherwise sorting is required.
The changes for the 2)check above has gone into FromBaseTable.nextAccessPath(Optimizer optimizer, OptimizablePredicateList predList, RowOrdering rowOrdering) The implementation of this change requires us to see if the order by column is involved in equijoin with outer optimizable's indexed column. If yes, then we know that since outer optimizable is ordered, the rows qualified via the equijoin will also be ordered and hence sorting can be avoided. But if this is not true, then we can't rely on outer optimizables' rows to be ordered on the order by column. To avoid sorting, we need to identify this case 2) as another case when the column can be marked as always ordered and that is when there is an equijoin predicate on the order by column with some other column
which is already known to be always ordered. Taking the query from wisconsin as an example will explain this behavior
select * from --DERBY-PROPERTIES joinOrder=FIXED
TENKTUP1 -- DERBY-PROPERTIES index=TK1UNIQUE1
, TENKTUP2 -- DERBY-PROPERTIES index=TK2UNIQUE1
where TENKTUP1.unique1 = TENKTUP2.unique1
order by TENKTUP1.unique1, TENKTUP2.unique1;
For the above query, as per the current trunk codeline, none of the order by columns are marked as always ordered because there is no constant comparison predicate on them. But, for the given join order, with TENKTUP1 as the outermost resultset and with the index TK1UNIQUE1, we know that the current row ordering at this point is going to ensure that rows from TENKTUP1 are ordered on UNIQUE1. Next, when we process TEKTUP2 in the 2nd join order position, we find that there is no constant predicate on TENKTUP2.unique1 and hence we conculde that the rows from TENKTUP2 are not going to be ordered and we decide to force a sort node on the top of the query. But in reality, even though the outer optimizable is not a single row resultset, it is ordered on TENKTUP1.unique1 and hence all those rows from outer optimizable are going to be ordered on TENKTUP1.unique1 and the inner optimizable has an equality join on
TENKTUP1.unique1 using the order by column TENKTUP2.unique1 What that translates to is that even if there will be multiple scans into TENKTUP2, the rows qualified are going to be all ordered because of the equijoin between the outer and inner optimizables on the order by columns. So, with my latest patch, I have expanded the notion of always ordered columns to include both constant comparison predicates AND ordered column that has equijoin with an outer optimizable's ordered column.
I think this patch is also improving the existing queries to include a better path than what it was picking up before. Following is an example of one such query from wisconsin.
select * from TENKTUP1, TENKTUP2
where TENKTUP1.unique1 = TENKTUP2.unique1
and TENKTUP2.unique1 < 100
order by TENKTUP1.unique1;
For this query, the trunk currently decides to use TENKTUP1 as the outermost optimizable using the TK1UNIQUE1 index and then those rows are filtered using TENKTUP2.unique1 < 100. Each of the 2 tables involved in the query have 10000 rows each. So we are going through 10000 qualified indexed rows from TENKTUP1 and then applying TENKTUP2.unique1 < 100 on them. With the attached patch, we use TENKTUP2 as the outermost optimizable with the index TK2UNIQUE1 and only gets the indexed rows which satisfy TENKTUP2.unique1 < 100 and then on them, we use the equlity join to fetch qualified rows from TENKTUP1.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@783168 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/sql/compile/RequiredRowOrdering.java",
"hunks": [
{
"added": [
"import org.apache.derby.impl.sql.compile.PredicateList;"
],
"header": "@@ -24,6 +24,7 @@ package org.apache.derby.iapi.sql.compile;",
"removed": []
... |
derby-DERBY-3926-397f968f | DERBY-4331
Fixes a number of sort avoidance bugs that were introduced by the fix
for DERBY-3926. This check in backs out the equi-join part of the DERBY-3926.
The changes for this were isolated and were the only changes to
FromBaseTable.java. Backing out only this part of the 3926 checkin fixes new
problems identified in DERBY-4331, and continues to fix the problem queries in
DERBY-3926. Knowledge of equijoin is no longer used as a factor for
sort avoidance.
Also included is an update to the wisconsin tests. A number of diffs
resulted from different join order to maintain a sort avoidance plan. 2
queries identified in DERBY-4339 no longer use sort avoidance.
The new test cases were reported as part of 4331 were added to the
OrderByAndSortAvoidance test.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@801481 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/compile/FromBaseTable.java",
"hunks": [
{
"added": [],
"header": "@@ -462,13 +462,6 @@ public class FromBaseTable extends FromTable",
"removed": [
"\t\t\t\t\t//Check if the order by column has equijoin on another ",
... |
derby-DERBY-3926-8f4810aa | DERBY-3926
Tars Joris contributed a new reproducible script(test-script.zip) which has been granted to ASF for
inclusion. I changed the junit test which I added last week to be based on this new script. The changes
involved were data change only. The actual test cases in the junit test did not require any changes.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@769147 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3931-31887565 | DERBY-3931: Make generated columns tests independent so that they will run smoothly on platforms which run the test cases in reverse alphabetical order.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@709161 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3932-9a9b9326 | DERBY-3932: Add basic tests of permissions on generated columns.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@713158 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3934-4987173e | DERBY-3934: Removed unused method 'readUnsignedShort' from UTF8Reader.
Patch file: derby-3934-6a-UTF8Reader_remove_method.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@742380 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/jdbc/UTF8Reader.java",
"hunks": [
{
"added": [],
"header": "@@ -592,25 +592,6 @@ readChars:",
"removed": [
" /**",
" * Decode the length encoded in the stream.",
" * ",
" * This ... |
derby-DERBY-3934-71dca8c7 | DERBY-3934: Improve performance of reading modified Clobs.
There are two major changes with this patch:
a) The implementation of getInternalReader improves the performance of
getSubString significantly. This benefits the embedded driver too, but it
is crucial for the performance of all read operations on Clob from the
client driver.
Again, the mechanism used to get better performance is to keep an internal
reader around to avoid repositioning on every request.
b) Added caching and updating of the Clob character length.
Also added some tests for verifying that the Clob length is handled correctly.
Patch file: derby-3934-4a-getinternalreader_cachedlength.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@726683 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/jdbc/TemporaryClob.java",
"hunks": [
{
"added": [
"import java.io.FilterReader;"
],
"header": "@@ -24,6 +24,7 @@ package org.apache.derby.impl.jdbc;",
"removed": []
},
{
"added": [
"imp... |
derby-DERBY-3934-910b77f0 | DERBY-3934 (partial): Improve performance of reading modified Clobs.
The following files are touched (all in derby.impl.jdbc):
*** EmbedClob.
Updated call to ClobUpdatableReader. The change of the position argument is
intentional.
*** TemporaryClob
Replaced the ClobUpdatableReader returned by getReader with a UTF8Reader.
Internal handling of TemporaryClob should deal with changing contents
specifically, or create a ClobUpdatableReader where required.
Note also the use of the new CharacterStreamDescriptor class. This piece of
code will probably be changed later on, when there is more information about
the stream available. For instance, caching byte/char positions allows to skip
directly to the byte position through the underlying file API. This way, we
don't have to decode all the raw bytes to skip the correct number of chars.
*** ClobUpdatableReader
More or less rewritten. It now uses the new methods exposed by InternalClob to
detect changes in the underlying Clob content. Note that this class doesn't
handle repositioning, only detection of changes and forwarding of read/skip
calls.
Note the lazy initialization of the underlying reader.
WARNING: There is one thing missing, which is proper synchronization. Access to
store will be synchronized in other locations, but this class is not thread
safe. I haven't decided yet whether to synchronize on the reader object or the
root connection. I think the latter is the best choice. Does anyone know
anything about the cost of taking locks on the same object multiple times?
*** StoreStreamClob
Replaced old UTF8Reader constructor with the new one. Again, this code needs
to be updated when more information about the stream is available. This is to
allow UTF8Reader to perform better.
*** UTF8Reader
Added a new constructor, using the new CharacterStreamDescriptor class.
Removed one constructor.
Retrofitted the second old constructor to use CharacterStreamDescriptor. This
will be removed when the calling code has been updated.
The old method calculating the buffer size will also be removed.
Stopped referencing PositionedStoreStream, using PositionedStream interface
instead. This allows the positioning logic to be used for both store streams
and LOBInputStream streams.
The reader has been prepared to be able to deal with multiple data offsets,
i.e. handling several store stream formats. For instance, the current
implementations has an offset of two bytes, where as the planned new one will
have an offset of at least five bytes. LOBInputStream has an offset of zero
bytes (no header information).
From now on, position aware streams are not closed as early as before, because
we might have go backwards in the stream. Streams that can only move forwards
are closed as soon as possible (as before).
Note that this patch doesn't fix the most serious performance issue. This will
be done in a follow-up patch by implementing getInternalReader in TemporaryClob.
Patch file: derby-3934-3a-clobupdreader_utf8reader.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@724294 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/jdbc/StoreStreamClob.java",
"hunks": [
{
"added": [
"import org.apache.derby.iapi.jdbc.CharacterStreamDescriptor;"
],
"header": "@@ -32,6 +32,7 @@ import java.io.Writer;",
"removed": []
},
{
"add... |
derby-DERBY-3934-9b9c25ad | DERBY-3934: Added two tests for Clob modifications (character replacement).
Patch file: derby-3934-1a-clob_replace_test.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@720767 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3934-a694c19f | DERBY-3934: Improve performance of reading modified Clobs.
Removed deprecated constructor and adjusted calling code.
Patch file: derby-3934-5a-UTF8Reader_cleanup.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@736000 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/jdbc/EmbedCallableStatement20.java",
"hunks": [
{
"added": [],
"header": "@@ -35,7 +35,6 @@ import java.sql.Blob;",
"removed": [
"import java.net.URL;"
]
},
{
"added": [
"import org.apa... |
derby-DERBY-3934-d9319b8e | DERBY-3934: Improve performance of reading modified Clobs.
Added two new methods to InternalClob;
- getUpdateCount
- isReleased
These methods can be used to detect if the contents have been changed, for instance by streams reading from the internal Clob representation.
Patch file: derby-3934-2a-intclob_new_methods.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@721162 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/jdbc/InternalClob.java",
"hunks": [
{
"added": [
" /**",
" * Returns the update count of the Clob.",
" * <p>",
" * The update count is increased each time a modification of the Clob",
... |
derby-DERBY-3941-081a08c2 | DERBY-3941: Clean up import statements in StoredPage
Contributed by Yun Lee.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@765943 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/store/raw/data/StoredPage.java",
"hunks": [
{
"added": [
"import java.io.ByteArrayInputStream;",
"import java.io.ByteArrayOutputStream;",
"import java.io.EOFException;",
"import java.io.IOException;",
... |
derby-DERBY-3941-a5d378d2 | DERBY-3941: Unsafe use of DataInput.skipBytes()
Replaced calls to DataInput.skipBytes() with new utility method
DataInputUtil.skipFully().
Patch contributed by Yun Lee.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@766163 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/services/classfile/ClassInvestigator.java",
"hunks": [
{
"added": [
"import java.io.IOException;",
"import java.util.Collections;",
"",
"import org.apache.derby.iapi.services.io.DataInputUtil;"
],
... |
derby-DERBY-3944-06ac9fb4 | DERBY-3944: Always compile CHECK constraints in the schema of the target table.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@964402 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/compile/DMLModStatementNode.java",
"hunks": [
{
"added": [
" CompilerContext compilerContext = getCompilerContext();",
" ",
"\t\tcompilerContext.pushCurrentPrivType( Authorizer.NULL_PRIV);"
],
... |
derby-DERBY-3945-01aa1762 | DERBY-3945: Resolve unqualified function names in generation clauses against the current schema from DDL rather than DML time.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@719123 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/catalog/types/DefaultInfoImpl.java",
"hunks": [
{
"added": [
" private String originalCurrentSchema;"
],
"header": "@@ -56,6 +56,7 @@ public class DefaultInfoImpl implements DefaultInfo, Formatable",
... |
derby-DERBY-3947-e594ab0b | DERBY-3947: Cannot insert 994 char string into indexed column
A table created with "CREATE TABLE t (x varchar(1000) primary key)" could
encounter problems when a particularly long value of "x" was inserted,
because the index that was automatically created to support the PRIMARY KEY
constraint was created with a small page size. Such an insert statement
would get an error like: "Limitation: Record of a btree secondary index
cannot be updated or inserted due to lack of space on the page."
This change enhances TableElementList so that, when creating an index for
a constraint, it now checks the approximate length of the columns in the
index, and, if they are sufficiently long, automatically chooses a larger
default page size for the index conglomerate.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@886162 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/compile/TableElementList.java",
"hunks": [
{
"added": [
"import org.apache.derby.iapi.reference.Property;",
"import org.apache.derby.iapi.services.property.PropertyUtil;"
],
"header": "@@ -25,6 +25,8 @@ import... |
derby-DERBY-3948-6a17f800 | DERBY-534: Support use of the WHEN clause in CREATE TRIGGER statements
Reject references to generated columns in the NEW transition variables
of BEFORE triggers, as required by the SQL standard. See also
DERBY-3948.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@1527489 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3948-e7354480 | DERBY-3948: Forbid references to generated columns in the NEW variable of BEFORE triggers.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@718707 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/compile/CreateTriggerNode.java",
"hunks": [
{
"added": [
"import org.apache.derby.iapi.sql.dictionary.ColumnDescriptorList;"
],
"header": "@@ -34,6 +34,7 @@ import org.apache.derby.iapi.sql.compile.CompilerContext;",
... |
derby-DERBY-395-22ccbb42 | DERBY-395 Server-side "trace on" and "trace off" commands do not appear to be working correctly.
Contributed by Bryan Pendleton
Attached is a proposed fix. derbyall passed. I put a small comment in the code. I didn't add any new tests, which is unfortunate, but I didn't have any brilliant inside about an easy way to add such tests. Testing with the server tracing is already somewhat of a PITA because the server tracing interacts poorly with things like the security manager.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@373291 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/drda/org/apache/derby/impl/drda/ClientThread.java",
"hunks": [
{
"added": [],
"header": "@@ -30,8 +30,6 @@ final class ClientThread extends Thread {",
"removed": [
"\tprivate String traceDir;",
"\tprivate boolean traceAll;"
]
},... |
derby-DERBY-3950-97a8b1c7 | DERBY-3950: Prevent driving SELECTs from overriding the values of generated columns.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@718381 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/compile/ResultColumn.java",
"hunks": [
{
"added": [
" private boolean wasDefault;"
],
"header": "@@ -88,6 +88,7 @@ public class ResultColumn extends ValueNode",
"removed": []
},
{
"added":... |
derby-DERBY-3955-1b72b60b | DERBY-3955; test lang/selectivity.sql can be revived
committing patch 3 - which adds the remaining test cases from selectivity.sql to SelectivityTest.java
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@1560247 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3956-f9cb8886 | DERBY-3956: Remove method TemplateRow.checkPartialColumnTypes
Deleted the method, which always returned true and was only called from withing debug blocks (sane builds).
Patch file: derby-3956-1a.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@719008 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/store/access/btree/BTreeScan.java",
"hunks": [
{
"added": [],
"header": "@@ -1541,10 +1541,6 @@ public abstract class BTreeScan extends OpenBTree implements ScanManager",
"removed": [
" ",
" ... |
derby-DERBY-3964-03972928 | DERBY-3964: Fix NPE in evaluation of generated columns while processing an ON DELETE SET NULL referential action.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@722623 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/compile/DMLModStatementNode.java",
"hunks": [
{
"added": [],
"header": "@@ -575,7 +575,6 @@ abstract class DMLModStatementNode extends DMLStatementNode",
"removed": [
" "
]
},
... |
derby-DERBY-3966-99494f14 | DERBY-3966: Make 1.4 JDK optional when building Derby.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@726092 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/build/org/apache/derbyPreBuild/PropertySetter.java",
"hunks": [
{
"added": [
" if ( j14lib != null ) { setClasspathFromLib(J14CLASSPATH, j14lib, true ); }",
" if ( j15lib != null ) { setClasspathFromLib(J15CLASSPATH, j15lib, true ); }"
... |
derby-DERBY-3969-b8b524c0 | DERBY-3969: Fix NPEs when declaring constraints on generated columns without explicit datatypes.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@723184 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/sql/compile/ResultColumnList.java",
"hunks": [
{
"added": [
"\t{",
" return getResultColumn( columnName, true );",
"\t}",
"",
"\t/**",
"\t * Get a ResultColumn that matches the speci... |
derby-DERBY-3970-dd2650ff | DERBY-3970: PositionedStoreStream doesn't initialize itself properly.
Makes PositionedStoreStream initialize itself properly by calling initStream and resetStream on the underlying Resetable.
Patch file: derby-3970-1a-PositionedStoreStream_init.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@722812 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/jdbc/EmbedBlob.java",
"hunks": [
{
"added": [
" myStream = new PositionedStoreStream(dvdStream);"
],
"header": "@@ -195,18 +195,8 @@ final class EmbedBlob extends ConnectionChild implements Blob, EngineLOB",
... |
derby-DERBY-3972-7b9c4ca8 | DERBY-3972; patch 3 modifies 2 tests to optionally take a property for
a different initial context factory than sun's
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@741227 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3975-92941262 | DERBY-3975: SELECT DISTINCT may return duplicates with territory-based collation
Made the implementation of hashCode() in the collation-sensitive
subclasses of SQLChar consistent with the collation-sensitive
implementations of equals(), compareTo() and stringCompare().
Also extended CollationTest with tests for SELECT DISTINCT on strings
that contain different characters but are considered equal in French
locale.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@728822 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/types/SQLChar.java",
"hunks": [
{
"added": [
" if (SanityManager.DEBUG) {",
" SanityManager.ASSERT(!(this instanceof CollationElementsInterface),",
" \"SQLChar.hashCode() does not wo... |
derby-DERBY-3977-4d20e641 | DERBY-3977: Clob.truncate with a value greater than the Clob length raises different exceptions in embedded and client driver.
Changed the exception thrown in the embedded driver when calling Clob.truncate with a length greater than the Clob length from XJ076 to XJ079 (see also release note).
The client and the embedded driver behavior is now consistent.
Updated tests.
Patch file: derby-3977-1a-change_emb_exception.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@726695 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3978-a63282c4 | DERBY-3978: Clob.truncate(long) in the client driver doesn't update the cached Clob length.
Make the client driver update the length of the Clob when it has been truncated.Added a few tests.
Patch file: derby-3978-1a-regression_tests.diff, derby-3978-2a-update_length.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@724657 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3980-11850ac0 | DERBY-3980 Conflicting select then update with REPEATABLE_READ gives lock timeout instead of deadlock
Just javadoc changes and adding a new test for the issue that can't be enabled until we have a fix. This check-in does not fix the issue in any way.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@726121 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/services/locks/Deadlock.java",
"hunks": [
{
"added": [
"\t * Walk through the graph of all locks and search for cycles among",
"\t * the waiting lock requests which would indicate a deadlock. A simple",
"\t * deadlock c... |
derby-DERBY-3980-e0699eac | DERBY-3980: Conflicting select then update with REPEATABLE_READ gives lock timeout instead of deadlock
DERBY-5073: Derby deadlocks without recourse on simultaneous correlated subqueries
Added more comments describing the deadlock detection algorithm.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@1084561 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/services/locks/Deadlock.java",
"hunks": [
{
"added": [
" * <p>",
" * Code to support deadlock detection.",
" * </p>",
" *",
" * <p>",
" * This class implements deadlock detection by searchi... |
derby-DERBY-3981-1aa5b64d | DERBY-3981: Improve distribution of hash codes in SQLBinary and SQLChar
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@731929 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/iapi/types/SQLChar.java",
"hunks": [
{
"added": [
" /**",
" * The pad character (space).",
" */",
" private static final char PAD = '\\u0020';",
""
],
"header": "@@ -112,6 +1... |
derby-DERBY-3981-dd584838 | DERBY-3981: Improve distribution of hash codes in SQLBinary and SQLChar
Added a performance test which shows the worst-case behaviour of the
current hash function. The test performs SELECT DISTINCT on a CHAR(5)
column and on a CHAR(5) FOR BIT DATA column. The current hash function
maps all the rows in the test to the same bucket and the hash table
used for elimination of duplicates becomes a linked list.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@730689 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3982-daa4827a | DERBY-3982: Commit Ole's DERBY-3982_p2_diff.txt patch, which makes it easy to specify patch releases as starting points for the upgrade tests.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@728024 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3988-636e8e5f | DERBY-3988: Second attempt to always build the JDBC4 support when compiling Derby.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@728693 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/build/org/apache/derbyPreBuild/PropertySetter.java",
"hunks": [
{
"added": [
" private static final String J16LIB = \"j16lib\";",
" private static final String J16CLASSPATH = \"java16compile.classpath\";"
],
"header": "@@ -96,6 ... |
derby-DERBY-3989-a8132ce4 | DERBY-3989 / DERBY-4699
Made PropertySetter ignore Java 6 libraries if a Java 5 compiler is used.
If j16lib is specified explicitly in such an environment, the build will be
aborted (an error message will be displayed to the user).
Patch file: derby-3989-02-aa-dontUseJava6LibsWithJava5Compiler.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@954421 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/build/org/apache/derbyPreBuild/PropertySetter.java",
"hunks": [
{
"added": [
" if ( j14lib != null ) {",
" debug(\"'j14lib' explicitly set to '\" + j14lib + \"'\");",
" setClasspathFromLib(J14CLASSPATH, j14l... |
derby-DERBY-3989-d6b04208 | DERBY-3989: Allow the build to succeed on machines which have a Java 6 environment but not a Java 5 environment.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@734242 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/build/org/apache/derbyPreBuild/PropertySetter.java",
"hunks": [
{
"added": [
" * <li>java16compile.classpath</li>"
],
"header": "@@ -46,6 +46,7 @@ import org.apache.tools.ant.taskdefs.Property;",
"removed": []
},
{
"added": ... |
derby-DERBY-3991-979d9e84 | DERBY-3991: Clob.truncate(0) throws exception.
Allows a Clob to be truncated to a length of zero characters (empty string).
Patch contributed by Yun Lee <yun.lee.bj@gmail.com>.
Patch file: derby-3991-3.diff
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@764800 13f79535-47bb-0310-9956-ffa450edef68
| [] |
derby-DERBY-3993-8d23f446 | DERBY-3993 With IBM 1.6 T_RawStoreFactory fails with There should be 0 observers, but we still have 1 observers on Win 2K
The problem will only show up in SANE builds as that is the only time we do
the sanity check.
Xact.doComplete() is called near the end of the transaction to take care of any
cleanup prior to committing or aborting the transaction.
It calls notifyObservers(commitOrAbort) and it expects on return that each
observer has been notified, and all the observers are coded
to delete themselves from the observer list as part of this process. It then
asserts that the list should be empty on return.
The problem is that one of the DropOnCommit observer as part of it's
processing manages to add another observer to the list. I am guessing that the
problem becomes intermittent because either different
JVM's/memory layouts/hash algorithms
result in the order processing of the observer list to be different,
or different implementations handle the adding of an observer to the list
while scanning the list differently. There is nothing in the Observable
javadoc that guarantees and order or says anything about expected behavior
of notifyObservers() execution if another observer is added during the
execution so I don't think it is a jvm bug.
In my case in order to process a drop of a container marked drop on commit the
raw store interface requires it to first be opened. The code adds
a TruncateOnCommit as part of this open as that layer of the code does not
know why it is being opened. I believe it is this "new" TruncateOnCommit
observer which is left on the observer queue. Adding an extra notify to the
drop on commit processing seems to fix the unit test, I'll see if that causes
any problems in the full set of tests.
git-svn-id: https://svn.apache.org/repos/asf/db/derby/code/trunk@1082197 13f79535-47bb-0310-9956-ffa450edef68
| [
{
"file": "java/engine/org/apache/derby/impl/store/raw/data/DropOnCommit.java",
"hunks": [
{
"added": [
"",
"\t\t\t\tSanityManager.THROWASSERT(\"still on observer list \" + this);",
"\t\tif (arg.equals(RawTransaction.COMMIT) || ",
" arg.equals... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.