Re: [HACKERS] Open 6.3 issues

Bruce Momjian wrote:
>
> ORDER BY NULLs problem?

Hadn't time to look, yet...

> optimizer memory exhaustion with many OR's

I reported that it can't be fixed for 6.3 - please, move it into
TODO now. Nevertheless, I could apply patch to prepqual.c to
free memory whenever it's possible...

> Is GROUP BY duplicates fixed?

I got two reports that current code is OK - remove it.

> Do we have a self-join optimizer performance problem?

Yes. Not sure that I'll have time...

> Do we have a problem with GROUP BY without ORDER BY?
                                     ^^^^^^^
Could you remind what the problem is ?

> Problem with tables >2Gb

I still suggest to add elog(ERROR) to mdextend()...

> Can we improve vacuum locking issues?

I would leave vacuum as is. Improving of vacuum was on my 6.3 TODO
list - it's 6.4 TODO now :)

> 'Can not write block blind' error on createdb and regression collision

I'm fixing buffer manager now...

> subselect issues, NOT IN (empty query)

Closed: NOT IN (empty) and all others Op ALL (empty) now
returns TRUE.

> Check triggers regression test(Vadim)

Fixed already.

> Check select_views regression test

Results can't be identical on all platforms => change second query with

SELECT name, #thepath FROM iexit ORDER BY 1, 2;

Vadim