Bruce Momjian <maillist@candle.pha.pa.us> writes:
>> Has anyone thought of putting a bzip2-compressed tarball up there?
>> Might save bandwidth...
> I don't even know what that is.
I know what it is, and I also know that it has achieved near-zero
market penetration. Yes, it compresses better than gzip; but evidently
not enough better to persuade people to switch.
Another consideration you have to pay attention to in today's world is
patent status. gzip has stood the test of time and is widely agreed to
be patent-free. (It'd be pretty hard for anyone to secure a patent on
gzip at this late date, even though the cluelessness of the USPTO is
nearly unbounded.) bzip2's author claims it is patent-free, but that
really only means that *he* didn't patent it. I don't think anyone has
done a serious patent search on Burrows-Wheeler methods.
Eventually something will come along that's enough better than gzip
to warrant a universal upgrade cycle, but as far as I can see bzip2
ain't it. In any case I see no need for Postgres to be out front of
the curve on this question...
regards, tom lane
PS: If you want more info see http://www.faqs.org/faqs/compression-faq/,
item 78.