Re: snapshot too old, configured by time

Tom Lane wrote:

> After struggling with back-patching a GIN bug fix, I wish to offer up the
> considered opinion that this was an impressively bad idea. It's inserted
> 450 or so pain points for back-patching, which we will have to deal with
> for the next five years. Moreover, I do not believe that it will do a
> damn thing for ensuring that future calls of BufferGetPage think about
> what to do; they'll most likely be copied-and-pasted from nearby calls,
> just as people have always done. With luck, the nearby calls will have
> the right semantics, but this change won't help very much at all if they
> don't.

I disagree. A developer that sees an unadorned BufferGetPage() call
doesn't stop to think twice about whether they need to add a snapshot
test. Many reviewers will miss the necessary addition also. A
developer that sees BufferGetPage(NO_SNAPSHOT_TEST) will at least
consider the idea that the flag might be right; if that developer
doesn't think about it, some reviewer may notice a new call with the
flag and consider the idea that the flag may be wrong.

I understand the backpatching pain argument, but my opinion was the
contrary of yours even so.

--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services