Re: Unexpected interval comparison

Kyotaro HORIGUCHI <horiguchi(dot)kyotaro(at)lab(dot)ntt(dot)co(dot)jp> writes:
> By the way the adt directory is, as suggested by the name,
> storing files with names of SQL data types so "int128.c" among
> then seems incongruous. Is "int128_test.c" acceptable? int16.c
> will be placed there in case we support int16 or hugeint on SQL.

After further reflection I've decided to put int128.h in
src/include/common/, thinking that maybe someday it will be useful
on client side too. Also I've changed the test harness file to
be src/tools/testint128.c, so that it won't be confused with code
meant to be part of the backend.

> Back to 9.5 seems reasonable to me.

I poked around and noticed that before 9.4, we did not attempt
to guard against overflows in interval calculations at all.
So backpatch to 9.4 seems pretty defensible. The non-HAVE_INT128
code works fine in 9.4.

I've just about finished adjusting the patch for the back
branches, and will push in a little bit.

regards, tom lane