At 02:51 PM 5/20/99 +1000, Simon Burge wrote:

>I'm sitting on some hacks for gzip which can decode bzip2 files as well.
>There'd be nothing to stop people from using that "funny" version of
>gzip, bzip2ing _all_ the binary sets but still calling them *.tgz files
>so nothing else (including sysinst) needs changing.  Unfortunately this
>won't help for 1.4 without a little bit of work as the install media has
>already been built.

If I understand you correctly, you're saying that these files would be
called .tgz, but wouldn't be uncompressible with a standard gzip. That
sounds like a bad idea to me - if they aren't *standard* gzip files, then
don't give them a name that implies they are. Otherwise you'll likely
confuse and piss off people who try to unzip them with normal gzip (eg,
people like me who want to upgrade onto their current system by untarring
with the old tar/gzip that's already on their system). Remember that not
everybody uses sysinst, or even uses the install diskette.

Why not just call the files <something>.tbz or something like that? It
would be unfamiliar to people, who would then hopefully read the README
file which told them they needed bzip2, but at least the mailing lists
wouldn't be deluged with "The distribution files are corrupt!" messages.

(BTW, why did we switch to the (afaik) "Linux-ism" of .tgz instead of
.tar.gz? Is the savings of 3 characters really worth it? This screws up the
mode stuff in my emacs that lets it read .tar.gz files (it does gzip mode
when it sees .gz, then does tarmode on the results when it sees .tar, but
since these two modes are separate from each other, there isn't an easy way
to make it understand .tgz - I end up just renaming the file instead of
trying to decode elisp).)