[Lurker-users] [feature request] Support Gzipped archives

[<ka...@he...>]

It would be nice if Lurker could support Gzipped archives.  The
problem might be that in order to access into the mbox.gz, you'd have
to decompress it every time.  I don't know if that's going to make it
too slow or resource hungry to use or not.

There's some work done with making gzipped tar file's rsyncable.  If
you search around on the net, you'll find things about it.  It was
that if you change the first byte of a tar, the rest of the tar.gz
changes relative to the original.  Thus, the rsync algorithm is almost
useless.  What they did is to patch gzip so that every so often it
clears some sort of compresser state and starts over, chunking the
file in a way deterministic by the contents of the tar, so that it is
now rsyncable -- that is, large chunks of slightly modified tars that
are then compressed will remain the same from one version to the next.

If you could compress the mbox and have gzip start over at each '^From
..' line, perhaps you could seek there?  I know very little about gzip
and am not the one to ask about this.  I wonder if something like that
is already part of zlib or not, to solve similar problems for
compressing other data files where you need to be able to index into
the compressed file and start decompressing to get a known (indexed)
portion of it?

-- 
No, one more time, I do NOT need your "leadership", sir.
Please just GET OUT OF MY WAY.                                .''`.
 Debian -- The blue collar Linux distribution.               : :' :
 <URL:http://www.debian.org/social_contract>		     `. `'