Re: How to make Debian more attractive for users
Josselin Mouette <joss@debian.org> writes:
> Le jeudi 22 juillet 2010 à 16:09 +1000, Ben Finney a écrit :
>> Russ Allbery <rra@debian.org> writes:
>>> This one [claim of Debian's libraries being out-of-date] always
>>> boggles me and makes me wonder if we should present Debian unstable or
>>> testing as the "typical" installation. Debian testing (and often
>>> Debian unstable) is more stable than the distributions with equivalent
>>> up-to-date libraries, and those distributions generally never offer
>>> anything remotely like Debian stable. (RHEL is considerably more
>>> unstable than Debian stable *and* has even older software, for
>>> example.)
>> Which of the above uses of “stable” refers to stability (“slow rate of
>> change”), and which refers to reliability (“high likelihood of working
>> when needed”)? Too many conversations conflate the two, and in this
>> case I think the distinction is important.
> Having already seen a major postfix upgrade which broke all existing
> installations in a RHEL update, I think we can talk of both stability
> and reliability in both contexts.
Amen. The experience for us with RHEL was kernel updates, but similar
principle.
Also, I think you can't have one without the other, really. It's very,
very hard to do reliable change of software, particularly when you're not
the institution writing the software or doing comprehensive QA (not that
there's much in the way of comprehensive QA in the free software world
other than lots of people running it and reporting bugs -- traditional QA
is usually too expensive for the benefit). I'm dubious that we'd ever
achieve the stability that Debian stable offers without the update policy
we use.
--
Russ Allbery (rra@debian.org) <http://www.eyrie.org/~eagle/>
Reply to: