Le mardi 22 janvier 2013 01:54:50, Michael Paquier a écrit : > On Tue, Jan 22, 2013 at 9:27 AM, Robert Haas <robertmh...@gmail.com> wrote: > > > On Mon, Jan 21, 2013 at 6:23 PM, Michael Paquier > > <michael.paqu...@gmail.com> wrote: > > > Yes, that is one of the most important patches in the list, and I could > > put > > > some effort in it for either review or coding. > > > > I think it would be great if you could elaborate on your reasons for > > feeling that this patch is particularly important. > > > Sure. recovery.conf has been initially created for PITR management, but > since 9.0 and the introduction of streaming replication it is being used > for too many things that it was first targeting for, like now it can be > used to define where a slave can connect to a root node, fetch the > archives, etc. I am seeing for a long time on hackers (2010?) that postgres > should make the move on giving up recovery.conf and merge it with > postgresql.conf. > > I didn't know about the existence of a patch aimed to merge the parameters > of postgresql.conf and recovery.conf, and, just by looking at the patch, > half of the work looks to be already done. I thought it might be worth to > at least update the patch or provide some feedback. > > I agree that this might break backward-compatibility and that it would be > more suited for a 10.0(?), but as 9.3 development is already close to its > end, progressing on this discussion and decide whether this could be > shipped for 9.3 or later release is important. If it is decided to give up > on this feature, well let's do that later. If it is worth the shot, let's > put some effort for it.
I though the idea is that for 9.3 we can have new feature, so everything can go in postgreql.conf, and also allows using recovery.conf so it does not break backward-compatibility. -- Cédric Villemain +33 (0)6 20 30 22 52 http://2ndQuadrant.fr/ PostgreSQL: Support 24x7 - Développement, Expertise et Formation
