Subject: Archiver comparisons

14. *****
 Q: What is the best archiver?

 A: Some netters have made accurate measurements of the packing /
unpacking times, and the size of the archives produced.  I won't try
to repeat those interesting experiments, but express some subjective
views instead. If you want to know more, the Usenet newsgroup
comp.compression is the one to read. Also see material like
ftp://garbo.uwasa.fi/pc/arcers/act-27.zip for the comparisons.
   If you are a heavy user, there is no choice. You'll need all the
archivers. (A good shell like shez will make life much easier for
you then).
   Let's start from Zoo. Personally, I found the old Zoo 2.01
prohibitively difficult to use beyond simple unpacking. Its
compression ratio was week. But it has been available for Unix and
MS-DOS for a long time, from the same author for both the platforms,
and this has made it the obvious choice for example for the
distribution binary postings in comp.binaries.ibm.pc despite what
used to be a poor ratio.
   The introduction of Zoo version 2.10 changed the situation quite
a lot for the better. Its compression is at good par with the other
methods, but it is still slow, and the commands are complicated, but
there is a good self-documentation to alleviate that. Furthermore,
there are problems with the file datestamping on Unix. I have
discussed potential further Zoo updates with the author Rahul Dhesi,
and he confirmed long ago that he was working on these aspects. One
has to be realistic though, since the last time around the updating
took over one year. Rahul is a busy man. It seems to be the same
with this promised update, also. Nothing whatsoever since.
   PKWare's PKZIP has become fairly dominating in bulletin boards
and many FTP sites. For example it is the only method we currently
accept in the uploads to our Garbo MS-DOS FTP archives. PKZIP makes
reasonably tight packages, but is slow in making them on old PCs.
With a 386 or 486 this does not matter any more. Personally, as an
MS-DOS FTP site moderator, I like the possibility of keeping the
date stamp of .zip archives in line with the date stamp of the
latest file within the package. With zipped archives this is very
easy. The attractiveness of zipping in archive maintenance in much
increased by Zip being available also for the Unix environment
courtesy of the Info-ZIP programming group. The support for the Unix
zip is unrivalled by the other archivers.
   PKWare almost blew it with its version 2 release. They advertized
version 2 on PC magazines for almost a year without it being
available, making it one of the best know cases of shareware
vaporware. On top of that the first 2 version, called 2.04C turned
out with far too many prohibitive bugs to be acceptable as a
compression method. For example Garbo archives refused to adopt
2.04C. PKWare then produced updates in rapid succession, naming them
versions 2.04E and 2.04G. The current 2.04G release has done very
well indeed, even if its advanced features occasionally cause some
consternation.
   The Japanese lharc is tight (comparable to, and often better than
PKZIP 1.10 was) and it is available for Unix, as well. But both
MS-DOS and Unix lharc totally lack user support (but who needs it
anyway for these facilities). The author is not even reachable. You
may experience difficulties with the path names when using lharc.
Since versions beyond 2.0 LHarc has been called LHa, and it is even
tighter than before. Lha can make self-extracting packages with very
little overhead, and is dominant among the compression archivers in
that respect.
   What about the old ARC method? PKWare's pkpak (originally called
pkarc) is fast in making archives, but its compression is not very
strong. But I still like it for some purposes because of its concise
one-line format of file comments, which I prefer to the multi-line
format used by pkzip and zoo.
   SEA's arc is available both for MS-DOS and Unix (from different
authors). But it has lost its former popularity and goodwill
(unfairly or not) among many users. First with the lawsuit with
PKWare (the author of pkarc aka pkpak, and pkzip), and subsequently
with its product development policy, which has left very much to be
desired, and caused some frustrated postings in the Usenet news. It
seems that SEA missed out on its golden opportunity to be the
dominant archiver. Now it is but vanishing from the scene. I have
seen postings referring to a new incompatible commercial version ARC
7.0, which is not to be distributed through bulletin boards and ftp
sites at all. I really fail to understand SEA's self-defeating
policies. SEAs new .arc is cannot be handled with the old versions,
but I have been told that /pc/arcers/arce*.zip can be used for
extraction.
   NoGate's Pak (as far as I know) is not available for Unix, only
for MS-DOS. But it can handle .arc, .pak, .sdn, and .zip formats,
and is the only one having this "multiple" property.
   Last, but certainly not least let's consider ARJ by Robert Jung.
ARJ, in version 2.50a at the time of last updating this item, is
turning out to be the most serious alternative to PKZIP. ARJ's
ability to span an archive on multiple disks is considered one of
its strong points. The author used to be active on the Usenet news,
and ARJ has thus good support. The lack of a proper Unix version is
however a major weakness. There only is unarj241.tar.Z. Some users
also complain about the switches being quite complicated to master.
You can, of course, alleviate that problem by using a shell like
SHEZ.
   There are also some other, tight methods like hpack, ha and sqz,
but mostly they have only a marginal following.
--------------------------------------------------------------------