I cc'b back to list, hope this is ok?
On Fri, 2002-12-13 at 14:56, george young wrote:
> On 13 Dec 2002 12:05:36 +0000
> Hannu Krosing <hannu@tm.ee> wrote:
>
> > On Fri, 2002-12-13 at 03:22, johnnnnnn wrote:
> >
> > In worst case you could generate the entries in pg_class table without
> > building the actual index and then drop or rollback when the explain is
> > ready.
> >
> --> Of course you could just determine all possibly useful indexes and <--
> --> generate then anyhow an then drop them if they were not used ;) <--
>
> Why not! At least for selects, this seems like the ideal. For insert
> and update, you have to deal with updating the superfluous indexes --
> does the planner include index updating in its work estimates?
Probably not - the work should be almost the same (modulo cached status
of index pages) for any plan.
At least I think we don't optimize the plan for different index access
patterns for updating indexes.
> > For queries
> that use functions in the where clause, you'd have to parse enough to know
> to include indexes on the functions (I know-- the last time I said "all I
> have to do is parse ..." I was really sorry later...).
--
Hannu Krosing <hannu@tm.ee>
