Re: Re: new patch of MERGE (merge_204) & a question about duplicated ctid

On Tue, Jan 04, 2011 at 09:27:10PM -0500, Greg Smith wrote:
> David Fetter wrote:
> >How about implementing an UPSERT command as "take the lock, do the
> >merge?" That way, we'd have both the simplicity for the simpler cases
> >and a way to relax consistency guarantees for those who would like to
> >do so.
>
> Main argument against is that path leads to a permanent non-standard
> wart to support forever, just to work around what should be a
> short-term problem. And I'm not sure whether reducing the goals to
> only this actually improves the ability to ship something in the
> near term too much.

I think I haven't communicated clearly what I'm suggesting, which is
that we ship with both an UPSERT and a MERGE, the former being ugly,
crude and simple, and the latter festooned with dire warnings about
isolation levels and locking.

If shipping with a "wart," as you term it, isn't acceptable, then I'd
advocate for going with just MERGE and documenting it inside and out,
including one or more clearly written UPSERT and/or REPLACE INTO
recipes.

Cheers,
David.
--
David Fetter <david(at)fetter(dot)org> http://fetter.org/
Phone: +1 415 235 3778 AIM: dfetter666 Yahoo!: dfetter
Skype: davidfetter XMPP: david(dot)fetter(at)gmail(dot)com
iCal: webcal://www.tripit.com/feed/ical/people/david74/tripit.ics

Remember to vote!
Consider donating to Postgres: http://www.postgresql.org/about/donate