On Tue, 2004-10-12 at 11:54, Sean Chittenden wrote:
> :-/ I have zero understanding or knowledge of the regression test
> suite.
It is trivial: add some SQL to src/test/regress/sql/foo.sql, run the
tests, hand-verify the output (e.g. by looking at regression.diffs), and
then copy the updated output from src/test/regress/results to
src/test/regress/expected
> Given the simplicity of the casts, does this really need a
> require a regression test?
Yeah, I suppose not. Personally I'd always err on the side of doing more
testing rather than less (I don't see much point in striving for
minimalism in a test suite), but you're right that this functionality is
probably sufficiently trivial it doesn't need tests.
-Neil