[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

problem with buildds == problems with pool space



        Hi!

 I noticed this nice behavior from the backports infrastructure:  an old
version of a package is only removed if its built on at least the same
architectures than the old one.  In an ideal world this is a really nice
and helpful approach to push changes sooner to the users.

 Though, we don't live in a real world.  There are various problems with
some build hosts, might it be outages, lazy admins not signing the build
files, whatever.  I don't want to blame any buildd admin here, it's a
voluntary job and there might always be other things that are more
important.

 But, the current situation is that we have for quite a lot packages at
least three different versions in the pool:  for firefox, for wesnoth,
for the linux kernel 2.6 it's even _4_ different versions lying around
in the pool.

 I consider this a really unconvenient situation.  Some might say, 19G
currently aren't that big of a problem, but I consider this also as a
disservice to the users because they might think that some of this
architectures are still supported where technically they aren't really.

 How about a deadline of let's say 4 weeks or so at a maximum for any
architecture to catch up with the build of a specific package, and if it
wasn't able to do so remove the old version of that specific package?
There is no need to remove the whole architecture -- thought if it is a
real problem there it will purge itself over the time anyway.

 Any thoughts?  I hope I don't step on anyone's toes with this
suggestion, I think it's the reasonable thing to do.

 Thanks.
Alfie
-- 
(o)_    ,---ooO--00--Ooo---.  FAQ for at.linux: http://alfie.ist.org/LinuxFAQ/
(>)_  ,'   Gerfried  Fuchs  `. GPG-Pub-Key: http://alfie.ist.org/key-alfie.asc
    ,'     <alfie@ist.org>    `.     GPG-Fingerprint: 3A53 DA5E 797E 9168 351F
    `- http://alfie.ist.org/ --'      1024D/EC152942  6666 314F 7A95 EC15 2942

Attachment: signature.asc
Description: Digital signature


Reply to: