Re: Extension Templates S03E11

On Sun, 2013-12-01 at 15:48 +0100, Dimitri Fontaine wrote:
> Jeff Davis <pgsql(at)j-davis(dot)com> writes:
> > I don't see why we are trying to accommodate a case where the author
> > doesn't offer enough full SQL scripts and offers broken downgrade
> > scripts; or why that case is different from offering broken upgrade
> > scripts.
>
> That's fair enough I guess. I will work on automating the choice of the
> first full script to use then, for next patch version.

Can we separate this feature out? It's an issue with extensions today,
and I'm eager to make some progress after the explosion of differing
opinions today.

Robert, do you think this is an acceptable approach to solve your pet
peeve here:

http://www.postgresql.org/message-id/CA
+Tgmoae3Qs4QbQfxOUzZFxRSxA0zy8ibSOYSuuTzDUMPeAkAg(at)mail(dot)gmail(dot)com

As an aside, I'm still not entirely sure why pg_dump omits the version
of the extension. It seems to introduce unnecessary ways to fail during
restore -- e.g. you've updated to (and depend on) 1.1, but restore will
try to restore using an old default of 1.0. Is it so that extension
authors can drop support for old SQL scripts and the restore will just
find the newer one?

If the reason the extension version is left out is because we can't find
the upgrade path to the requested version, this feature should fix that.

Regards,
Jeff Davis