Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!ucsd!swrinde!zaphod.mps.ohio-state.edu!sdd.hp.com!elroy.jpl.nasa.gov!ames!amelia!roelofs
From: roelofs@amelia.nas.nasa.gov (Cave Newt)
Newsgroups: comp.binaries.ibm.pc.d
Subject: Re: PKZIP 1.10 vs. LHARC
Message-ID: <7239@amelia.nas.nasa.gov>
Date: 10 Jul 90 05:33:21 GMT
References: <6782@amelia.nas.nasa.gov> <8762@crdgw1.crd.ge.com> <1990Jun20.032819.23081@uwasa.fi> <4929@uwm.edu>
Organization: University of Chicago
Lines: 47

In article <4929@uwm.edu> chad@csd4.csd.uwm.edu (D. Chadwick Gibbons) writes:

>	A policy of PKWare inc. is not to release its sources.  To
>effectively produce utilities for UNIX, you need just that--sources.  Phil
>explains quite well how to perform the ZIP operations, it just takes some
>soul to bite the bullet and produce the code.  As you know, plenty of
>un-zip utilities exist, just no create-zip utilities.

As I've just recently been discussing with Ge' Weijers of the Netherlands,
this "bullet" is more like an 80mm shell--duplicating the compression
algorithm for Imploding is highly non-trivial, as they say in the mathematics
business.  And while Phil gives detailed algorithms for UNcompressing the
various types of encoding, the compression end of things is much more sketchy.
I don't imagine a PKZIP clone could be written in less than several weeks,
and that's assuming full-time effort....  If, however, somebody would care 
to prove me wrong, I'll send 'em a Mrs. Fields chocolate chunk cookie. :)


In article <8762@crdgw1.crd.ge.com> (bill davidsen) writes:

> Since there is no PKZIP for UNIX, LHARC is a lot faster there. So are
> ARC and ZOO. Phil has chosen not to compete in those systems.

nulspace@eng.umd.edu (Michael Chang) writes:

>Believe it or not, lharc is not faster at all on unix systems, at least
>between unix systems and PCs with sufficient hard drive space and RAM.
>I checked the source code, and it is the same for both DOS and Unix.  I'm
>sure that lharc could be optimized to work faster on unix, I only wish
>someone would go do it.

The point of Bill's posting was that LHARC wins over PKZIP under Unix,
because PKZIP does not exist under Unix.  Nobody was comparing times across
operating systems--for one thing, the performance under Unix will depend
a great deal on how many other users are on the system; and then, you can
always toss LHARC in the background and go do something else under Unix,
so in a sense it's infinitely faster than DOS.  And best of all, Unix runs
on Crays and DOS doesn't:  I daresay LHARC and just about anything else 
you'd care to try would win hands down on an 8-processor Y-MP (assuming
you could get the damn thing to compile correctly).

Speaking of which:  has anybody succeeded in making mods to LHARC to allow
it to run under Unicos (64-bit long ints, 46-bit shorts)?  If so, please
send me e-mail.  I've gotten it to compile, but something is breaking and
I haven't yet discovered what.  Many thanks...

Greg