2008/12/28 Tom Lane <tgl@sss.pgh.pa.us>:
> I've spent quite a bit of time reviewing the window functions patch,
> and I think it is now ready to commit, other than the documentation
> (which I've not looked at yet at all). Attached is my current patch
> against HEAD, sans documentation. This incorporates the recently
> discussed aggregate-function API changes and support for tuplestore
> trimming. There's a number of things that could be improved yet:
> * we really ought to have some support for non-built-in
> window functions
> * I think the planner could be a bit smarter about when to
> sort or not
> * tuplestore_advance and related code really needs to be made
> more efficient; it didn't matter much before but it does now
> but I think these things can be worked on after the core patch is
> committed.
>
I ran the patch witouht any errors. Though it's trivial, I noticed
window_gettupleslot has to be fixed a bit.
diff src/backend/executor/nodeWindowAgg.c.orig
src/backend/executor/nodeWindowAgg.c
1445a1446,1449
> /* pos = -1 means special on spool_tuples(), so check it before */
> if (pos < 0)
> return false;
>
1449c1453
< if (pos >= winstate->spooled_rows || pos < 0)
---
> if (pos >= winstate->spooled_rows)
Regards,
--
Hitoshi Harada
