At 21:23 -0400 on 06/26/2006, Gerhard Postpischil wrote about Re:
Program Objects:
Paul Gilmartin wrote:
In a recent note, Rick Fochtman said:
as (possible) access to VSAM clusters of the indexed, sequential
and relative
I've learned that my compression algorithm can create
records longer than
the original records, so I'm working on a fix for that, as well.
There's no fix for that (think Pigeonhole Principle). You must be
prepared to tolerate it.
There is a proof somewhere (Knuth?) to the effect that for every
record that's compressible, there's one that will be longer.
Empirically I ran across this in the seventies with Wylbur, which
replaced repeated blanks by a single character, and inserted a count
byte every sixteen non-blanks. Later versions of Wylbur used a flag
bit in the record to indicate it wasn't compressed. Perhaps
something similar could be used in ARCHIVER (I haven't looked at the
source in some decades).
I seem to remember that Huffman Code Compression can NEVER create a
"Compressed" Record/File longer than the Uncompressed version (so
long as you consider only the payload and ignore the < 600 byte table
needed for the expansion process).
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
