Re: datetime and timespan deprecated

> Not here.  I seem to recall that there are several other "remove in the
> next release" compatibility hacks in gram.y, most of them of more than
> one release's standing.  Shall we put them all on a "this IS going away,
> this time for sure" hitlist?

Sure. I'm not recalling any particulars, other than the ODBC hack which
is already covered now in the driver itself afaict. Not sure if we ever
had a detailed test case specified to test that one anyway. Peter?

Oh, there is an SQL9x vs PostgreSQL issue wrt whether timestamp defaults
to "with time zone" or not. I have it default to "with time zone" for
this release for compatibility with previous releases, but we could
consider changing that in 7.3 per SQL9x spec on defaults. To make that
easy, pg_dump should do the right thing by explicitly specifying this on
output, and it seems to do that already. So we have a pretty seamless
path to 7.3 for this issue.
                   - Thomas