Jeff Hoffmann <jeff@remapcorp.com> writes:
> i'll try updating some of the dedicated tests (box.sql, circle.sql,
> geometry.sql, lseg.sql, path.sql, polygon.sql), but i'm not sure where
> testing the rtree indexes should go. i think other index types are
> tested in select.sql, but i'd probably put them in geometry.sql. does
> anybody care? is there someone that oversees the methods and
> organization of the regression tests or do people just throw in new
> tests when there's something new?
AFAIK we have no regression-test-meister (though we should). Do what
seems reasonable.
About the only stylistic thing I'd suggest is to try to avoid machine
dependent results. For example, the existing geometry.sql test causes
a lot of uninteresting comparison failures on many machines because of
small variations in roundoff error. So, if you can exercise a feature
using only exact-integral inputs and results, do it that way rather than
making up "realistic" test data. (A lot of people would be very happy
if you could revise this problem away in geometry.sql ... but if that
seems like more work than you bargained for, don't worry about it.
Extending the test coverage is the high-priority task, I think.)
I'd be inclined to agree that rtree indexes should be tested in one
of the existing geometry-related tests, or perhaps in a brand new
regression test, rather than sticking them into the generic select.sql
test.
Thanks for taking a shot at it!
regards, tom lane