Ühel kenal päeval, K, 2006-04-12 kell 10:29, kirjutas Jim C. Nasby:
> On Wed, Apr 12, 2006 at 04:53:20PM +0200, Thomas Hallgren wrote:
> > <NestedLoop cost="1.06..40.43" rows="5" width="244">
> > <JoinFilter publicTenk1Unique2="int4_tbl.f1">
> > <HashAggregate cost="1.06..1.11" rows="5" width="4"/>
> > </JoinFilter>
> > </NestedLoop>
>
> Well, the downside is that such a format means explain output is now
> twice as long.
You can place end tags differently
<NestedLoop cost="1.06..40.43" rows="5" width="244"> <JoinFilter left="publicTenk1Unique2" right="int4_tbl.f1">
<HashAggregatecost="1.06..1.11" rows="5" width="4"/></JoinFilter></NestedLoop>
> But I'd love to see something like that as an option.
Me too
> I'd also still like to see an SQL-parseable version as well, since I think
> there's applications for that.
>
> As for those who can't manage to post EXPLAIN ANALYZE to the list; as
> long as ANALYZE isn't the default I don't see how making a less
> human-readable version the default will solve anything, because we'll
> still perpetually be asking people for the output of EXPLAIN ANALYZE. If
> we want to increase the number of people who provide useful information
> in initial performance questions, the answer is to make the information
> about what to submit more prominent.
We could also default to printing a NOTICE at the end of EXPLAIN, which
tells users thus: "If you plan to post this output to pgsql-hackers
list, you better post result of EXPLAIN ANALYSE" :P
------------
Hannu
