Re: Backpatching of "Teach the regular expression functions to do case-insensitive matching"

Robert Haas <robertmhaas@gmail.com> writes:
> On Mon, May 9, 2011 at 10:39 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> No objection here, but how will we do that? �The regression tests are
>> designed to work in any locale/encoding, and would become significantly
>> less useful if they weren't.

> I'm just shooting from the hip here, but maybe we could have a
> separate (probably smaller) set of tests that are only designed to
> work in a limited range of locales and/or encodings.  I'm really
> pleased that we now have the src/test/isolation stuff, and I think
> some more auxilliary test suites would be quite excellent.  Even if
> people didn't always want to run every single one when doing things
> manually, the buildfarm certainly could.

Hmm.  We don't need new infrastructure like the isolation tests do,
so another subdirectory seems like overkill.  I am thinking about a new
target "installcheck-collations" in src/test/regress/GNUmakefile that
creates a UTF8-encoding database and runs a different test schedule than
the regular tests.

The problem we'd have is that there's no way (at present) to make such
a test pass on every platform.  Windows has its own set of locale names
(which initdb fails to install as collations anyway) and we also have
the problem that OS X can be counted on to get UTF8 sorting wrong.
(It might be okay for case-folding though; not sure.)  Possibly we could
just provide an alternate expected file for OS X, but I don't see a
decent workaround for Windows --- it would pretty much have to have its
very own test case.

Andrew, what kinds of options have we got for persuading the buildfarm
to run such tests only on a subset of platforms?
        regards, tom lane