Boszormenyi Zoltan <zb(at)cybertec(dot)at>, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Peter Eisentraut <peter_e(at)gmx(dot)net>, Hans-Juergen Schoenig <hs(at)cybertec(dot)at>, Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>
On Thu, Feb 11, 2010 at 2:04 PM, Alvaro Herrera
<alvherre(at)commandprompt(dot)com> wrote:
> Boszormenyi Zoltan escribió:
>> Robert Haas írta:
>> > ...
>> > OK, please change it.
>> >
>>
>> New patch is attached with the discussed changes.
>
> This looks all wrong. PORTAL_ONE_SELECT is now being passed through
> FillPortalStore, which runs it to completion, whereas it was previously
> passed via PortalRunSelect first, which has different semantics
> regarding the "count" arg.
>
> Also, even if that weren't wrong, FillPortalStore states at its header
> comment that it is only used for the other two cases (ONE_RETURNING and
> UTIL_SELECT), but now is being used for ONE_SELECT as well.
I was all prepared to admit that I hadn't actually looked at the patch
carefully enough, but I just looked at (and CVS HEAD) again and what
you've written here doesn't appear to describe what I'm seeing in the
code:
if ((portal->strategy != PORTAL_ONE_SELECT) && (!portal->holdStore))
FillPortalStore(portal, isTopLevel);
So one of us is confused... it may well be me.
...Robert
