Re: [HACKERS] timezone problem?

> Yes, and even worse, CST also is "China Standard Time" in some operating
> systems. I won't go into how broken every operating system is vis-a-vis
> Chinese timezones (but, believe me, it's a mess).
> >From here on out, I'm strictly in "+0800".
> I've become convinced that any project that thinks it is going to keep
> comprehensive, accurate, non-conflicting, non-obsolete timezone information
> in an application-specific table is woefully misguided.

Yup. And that brings up an issue: I would like to have the *default*
style for date/time output in 7.0 be ISO, rather than the current
"traditional Postgres". I was waiting for a major rev to do this (but
it probably should have happened before the y2k change of year). It's
a one-liner to update this.

Bruce, can you add this to the "critical items" for 7.0, barring fatal
objections from other developers?

> >btw, the patch also tries to fix the "GMT+hhmm" timezone format
> >reported recently as being available on FreeBSD; perhaps someone could
> >test that at the same time.
> Does this patch apply cleanly against 6.5.3?

I'm not certain, but it should since this area of the code does not
change very much. If you apply with

cd src/backend/utils/adt
patch < dt.c.patch

you should get a dt.c.orig so can revert easily if necessary.

- Thomas

--
Thomas Lockhart lockhart(at)alumni(dot)caltech(dot)edu
South Pasadena, California