"Jonah H. Harris" <jonah.harris@gmail.com> writes:
> we're working on a prototype to reduce WAL I/O and index updates in a
> large percentage of OLTP situations by employing an update-in-place
> under *safe* conditions.
AFAICS there are no circumstances, ever, in which update-in-place is
"safe". (No transaction can guarantee that it will commit.)
> In my opinion, I don't think we should have an option to allow the
> indexes to become corrupt.
Martin's proposal at least looks sensible; he just hasn't quite made the
case that it's worth doing. If you're running a system that hardly ever
crashes, you might be willing to accept index rebuilds during crash
recovery, especially for indexes on relatively small, but frequently
updated, tables (which should have reasonably short rebuild times).
Obviously this would have to be configurable per-index, or at least
per-table, and I agree that it likely would never be the default.
But it could be a good tradeoff for some cases.
regards, tom lane