Hello Mark!
20 Feb 97 20:44, Mark Madronio wrote to Hans Mangold:
 -=>> Quoting Hans Mangold to All <=-
 HM>> Some of you know that I am very excited about Robert Jung's
 HM>> (ARJ) new JAR compressor for DOS and for Windows 95 / NT.
 HM>> Take a close look at the results below; you might be VERY
 HM>> suprised with the results :-)
 MM> Hi, that's a nice table you had there. But they were all text files.
Mark, there was a reason behind my madness and it actually had nothing to do 
with *.txt files as I shall demonstrate later :-)  Inadvertently, I stumbled 
across a case where the entire theory of "solid" archivers is best 
demonstrated.  In essence, the old classics (ARJ, PKZip, LHA, etc.) compress 
file-by-file, looking for redundancy only within that one file.  "Solid" 
archivers, such as RAR and JAR, treat all files as one, so to speak, and look 
for redundancy across _all_ files.  If you looked at my example closely, 
there wasn't all that much difference using ARJ or Zip and JAR if I added the 
files one at a time.  Only once I archived all twelve very similar files at 
once did the "solid" archiving method really go to work.
 MM> Can you try doing it on maybe a 2Mb game or application so we could
 MM> see what it can do for program files? I'm particularly interested in
 MM> the results when using maximum compression and maximum dictionary size
 MM> for both RAR and JAR.
Mark, I'm sorry but I just don't have the time to do that at the moment.  
Take a look at the *.docs supplied with JAR; there are some recognized, 
more-or-less standardized benchmarks using different files.
See also next message.....
Cheers, Hans
... Be kind to animals - hug a SysOp!!!
--- GoldED/386 2.50+ / Squish / Maximus / Binkley / WINDOWS 95 / V34+
---------------