Обсуждение: SQL99 time zones

I'm working on patches to implement the (slightly brain damaged) SQL9x
time zone spec. This allows one to specify a numeric time offset for the
time zone.

The changes do not affect system catalogs, touching the date/time
routines in a few places and affecting gram.y and variable.c. To get the
range of choices to include those specified by the standard I am
changing the interface to SetPGVariable() to accept parser nodes rather
than just a string as an argument. This should make some of the other
SET variables easier to support too.

These changes are not quite ready to go (since I need to go back and fix
up the other parameters supported by SetPGVariable()) but I believe that
they are very low risk since the default behavior would stay the same.

I expect to have this ready at the beginning of next week. One might
consider this a bug fix...

Comments?
                       - Thomas

Thomas Lockhart <lockhart@fourpalms.org> writes:
> I'm working on patches to implement the (slightly brain damaged) SQL9x
> time zone spec. This allows one to specify a numeric time offset for the
> time zone.  [ ... ]
> Comments?

While this is doubtless a good thing, I'm starting to feel very itchy
about the fact that we've slipped beta a couple of weeks now while you
hack up "one more" datetime improvement.  We've got to have some closure
on this process.

At this point I guess the reasonable thing to do is say "Beta on Monday,
this time for sure!".  If you haven't committed these changes by Monday,
I think you should hold them over for 7.3.
        regards, tom lane