Josh Berkus wrote:
Neil, Robert:

"As for the "WAL recovery bug", AFAIK no such bug has been reported "in
the last few days". Exactly what issue are you referring to?"

That's my bug; I filed it on Wednesday.

However, it is not 100%; that is:
1) While Tom and I are pretty sure that the issue *could* cause the behavior reported, we're not completely certain that it *did*; i.e. in the two reported cases, one actually turned out to be something else, and the other could possibly be something else as well.

2) Nobody has tested that switching the order of those 2 lines in 7.2.3 doesn't cause any problems, to date.

I'm not saying that it's not potentially a patchable bug. We're just not ready to patch it yet.
Ok, this might not be such an important fix after all then? The wording of it at the time did make it sound important, but if it somehow has bad interactions we would be shooting ourselves in the foot with it.

Any guess-timates on it's safeness and whether it really would be beneficial?

But I do vote for a 7.2.4 just because I can't upgrade a lot of my clients to 7.3.1 safely and there are a few easy patches for 7.2.3.
Alternately, I would suggest an omnibus patch for the 7.2.3 source code so that we don't set a precedent for branching development.
An interesting thought here is to know if Red Hat fixed *all* of the known PostgreSQL security flaws for 7.2.3 with their latest security release. It would be interesting to see their code if they did so, but from Tom's previous comments it would have meant a real lot of work.

It's probably better to put out a 7.2.4 than an omnibus patch though, as it gives a better foundation for everyone working on 7.2.x to safely move to. From the viewpoint of "it takes more skill to patch than to compile".

Regards and best wishes,

Justin Clift

"My grandfather once told me that there are two kinds of people: those
who work and those who take the credit. He told me to try to be in the
first group; there was less competition there."
- Indira Gandhi

---------------------------(end of broadcast)---------------------------
TIP 4: Don't 'kill -9' the postmaster

Reply via email to