Re: PATCH: CITEXT 2.0 v3

On Jul 9, 2008, at 13:40, Alvaro Herrera wrote:

> The problem is that we're in the middle of a commitfest, so
> everybody is
> busy reviewing other patches (in theory at least).

Of course.

> One thing that jumps at me is pgTAP usage, as Zdenek said. I
> understand
> that it's neat and all that, but we can't include the tests because
> they
> won't run unless one installs pgTAP which seems a nonstarter. So if
> you
> want the tests in the repository along the rest of the stuff, they
> really should use pg_regress.

It does use pg_regress. The test just load the included pgtap.sql file
to get the tap functions, and then away they go. If you run `make
installcheck` it works.

> It's not even difficult to use. Have a look at contrib/ltree/sql and
> contrib/ltree/expected for examples.
>
> If you want to push for pgTAP in core, that's fine, but it's a
> separate
> discussion.

Agreed. I've sent a couple of messages in a thread about that, the
latest this morning.

> The other possibility being, of course, that you are proposing
> citext to
> live on pgFoundry.

I'm not, actually. I mean, I have an updated version for 8.3, but it'd
be quite a pita to maintain them both, since the api for lowercasing
text is so much simpler in 8.4.

Best,

David