...
> Ok, so how should things work, then? While I agree that SQL92's spec
> is awkward and limited, we'd need a pretty good argument for breaking
> standards. Oliver is already wearing me down in this regard.
Well, the standard sucks ;)
My reference on this is Date and Darwen (I think that Date used another
word than "sucks", but his meaning is clear), who reinforced my
suspicion that the SQL9x date/time folks were in an altered state when
they formulated the standard. The standard is inconsistant, incomplete,
and does not match common and essential usage for dates and times. Other
than that, it does a great job with dates and times :))
I won't try to defend the current PostgreSQL implementation as the way
it should always be, but it does have hundreds of hours of work in it to
get where it is. Backing off to "if it isn't in the standard, then kill
it" is a step backwards. I see more than a few more hours of work coming
with the unbelievably short sighted glibc breakage recently introduced.
I'm not subscribed to -sql, and would think that if the discussions have
evolved from "how do I do this" to "how *should* we do this" then the
discussion should move to -hackers. I'm not subscribed to every list,
and really can not keep up with the ones I am on now. I recently
subscribed to -general because design discussions seem to erupt there,
but will likely unsubscribe soon. And expect that design happens on
-hackers where it is intended.
- Thomas