text
stringlengths
0
99.6k
ARC calculates a two byte checksum for each file in an
archive. The checksum is calculated using the bytes of
the original file before they are passed on to the
compression routines. When you verify an archive, ARC
actually de-compresses each archive entry and
calculates a new checksum using the bytes passed to it
by the decompressor. Both checksums should be the same.
If they do not match, an error message is displayed
indicating that the archive may have been corrupted due
to a disk error or a transmission error during upload
or downloading.
Again the syntax is the same as that for ARC/X.
By now you must have noticed that ARC/X ARC/P and
ARC/V are all minor variations of the same thing.
ARC VERSION 2.20 PAGE - 27
4) MEMORY MAP
$033c-$03ff - cassette buffer. used by ARC
$0801-$0fff - not used
$1000-$4fff - work space for ARC/C and ARC/X
CRUNCH string table is stored here
$5000-$7fff - workspace for ARC and MOVE
commands only. All other commands
leave this area alone.
$8000-$8fff - used in 80 column version. (ROM)
$9000-$97ff - not used
$9800-$9fff - 80 column screen.
Not used in 40 column ARC
$a000-$bfff - program area. ARC, MOVE, and DIR
$c000-$cfff - program area. Editor commands.
(sys 12*4096 to enable ARC after
a KILL)
$d000-$ffff - work space for ARC/C and ARC/X
You may notice that there is a rather significant
jump in the amount of workspace that ARC needs to do
its job from previous versions of ARC (28K to be
exact). This is due to the CRUNCH routines, which are
rather demanding in terms of memory. If you have a
program in memory when archiving or dearchiving a file,
then it will almost certainly be clobbered by ARC.
When invoking the ARC command, BASICs pointers are
not changed in any way by ARC. You may have to type NEW
before running a program after using the ARC command.
(otherwise you may get an ?out of memory error.)
ARC VERSION 2.20 PAGE - 28
THEORY OF OPERATION
All methods of data compression take advantage of
redundancy of one form or another. Run-length coding is one
of the simplest, and often the most effective techniques.
Graphics files often contain long sequences of the same
byte. Zeros for blank space, or 255's for filled in space.
Run-length coding recodes these long sequences as shorter
control sequences. For example, a graphical image stored in
RAM may look something like the following if viewed with
the machine language monitor:
.:2000 00 00 00 00 00 00 00 00
.:2008 00 00 ff ff ff ff ff 00
.:2010 00 00 00 00 00 00 00 00
.:2018 a0 0b ff ff ff ff ff ff and so on....
This could be stored on disk as the sequence:
fe 00 0a fe ff 05 fe 00 09 a0 0b fe ff 06
The first byte ($fe) is a control character. When the
unsqueeze routine encounters a one of these it gets the next
two characters and interprets them as a character
identifier and a count. Thus the first 3 byte sequence is
interpreted as 10 zeros, the next 3 byte sequence as 5 ff's
and so on. When a character is not repeated, it is simply
coded directly to the output file. (the $a0 at $2018 above)
And so the above is squeezed from 32 bytes down to 14.
Database programs often sacrifice disk space in order
to gain speed. Relative files, for instance, store their
data at the beginning of each record, and pad the record
with zeros. Since every record is the same length, the DOS