Re: [PG13] Planning (time + buffers) data structure in explain plan (format text)

On Wed, 19 Aug 2020 at 21:05, Julien Rouhaud <rjuju123(at)gmail(dot)com> wrote:
>
> On Wed, Aug 19, 2020 at 08:49:48PM +1200, David Rowley wrote:
> > However, I'm not quite sure how we should handle if someone does:
> > EXPLAIN (BUFFERS on, SUMMARY off). Without the summary, there's no
> > place to print the buffers, which seems bad as they asked for buffers.
>
>
> But this won't be as much a problem if ANALYZE is asked, and having different
> behaviors isn't appealing. So maybe it's better to let people get what they
> asked for even if that's contradictory?

I'd say BUFFERS on, BUFFERS off is contradictory. I don't think
BUFFERS, SUMMARY OFF is. It's just that we show the buffer details for
the planner in the summary. Since "summary" is not exactly a word
that describes what you're asking EXPLAIN to do, I wouldn't blame
users if they got confused as to why their BUFFERS on request was not
displayed.

We do use errors for weird combinations already, e.g:

postgres=# explain (timing on) select * from t1 where a > 4000000;
ERROR: EXPLAIN option TIMING requires ANALYZE

so, maybe we can just error if analyze == off AND buffers == on AND
summary == off. We likely should pay attention to analyze there as it
seems perfectly fine to EXPLAIN (ANALYZE, BUFFERS, SUMMARY off). We
quite often do SUMMARY off for the regression tests... I think that
might have been why it was added in the first place.

David