Martijn van Oosterhout <kleptog@xxxxxxxxx> writes: > On Thu, Jul 12, 2007 at 09:04:38AM +0200, Hannes Dorbath wrote: >> This god like faith of some admins in package maintainers, that they >> know what's right, good and stable for them, sometimes really worries me. > The problem is the mismatch between what distrbuters want and what the > postgres team wants. For distributors "stable" means no behavioural > changes, whereas the postgresql team does bug fixes, some of which > definitly make behavioural changes that would make previously working > programs break. I think we have a pretty good track record of not doing that except when it's forced by a need to plug a security hole. However, distributors certainly have more constraints than one could wish. For instance, at Red Hat I can't just push a new Postgres update into RHEL releases at my whim --- there are company constraints based on available QA resources and suchlike. So sometimes the RHEL version of PG lags behind the community version just because of manpower/scheduling issues. They have been pretty good about letting me push security updates promptly, though. regards, tom lane
