"David E. Wheeler" <david@kineticode.com> writes:
> On Jul 13, 2008, at 10:19, Tom Lane wrote:
>> BTW, actually a better idea would be to put citext.sql at the front of
>> the list and just run the whole main regression series with it
>> present.
>> typ_sanity and oidjoins might possibly find issues too.
> Um, stupid question (sorry, I'm not familiar with how the regression
> tests are organized): serial_schedule doesn't look like a file into
> which I can insert an SQL file. How would I do that?
You're really making it into another test. Just copy the citext.sql
file into the sql/ subdirectory and add a "citext" entry to the schedule
file.
The last time I did this, I had to at least "touch" a corresponding
expected file (expected/citext.out) or pg_regress would go belly-up
without running the rest of the tests. There's no special need to
make the expected file match, of course, since you can just ignore
the "failure" report from that one test.
regards, tom lane