Tom Lane <tgl@sss.pgh.pa.us> writes:
> How exactly would you know whether the previous installation was built
> without HAVE_WCSTOMBS/HAVE_TOWLOWER? That's not exposed anywhere
> reliable. And it's not out of the question that somebody upgrading to
> a newer PG version might upgrade his OS too, so I would not think that
> checking what configure says now is trustworthy.
Oh, it's even worse than I imagined then. If you don't know where to
find the information, maybe the release notes should just propose to
REINDEX in any case?
> In general, though, this suggestion seems about two orders of magnitude
> more work than the particular case justifies. We might want to think
> about something like it for future releases, but it's not likely to
> happen for 9.3.
Yeah, I've been thinking that it's the case after sending the email
proposal. It might just be some accumulative effect that leads me to
that proposal. I still think that a tool able to asses if you're
concerned by "reindex gist indexes on that datatype" or suchlike would
be really good to have.
Regards,
--
Dimitri Fontaine
http://2ndQuadrant.fr PostgreSQL : Expertise, Formation et Support