Re: Changed SRF in targetlist handling

On Mon, Jun 6, 2016 at 2:31 PM, Alvaro Herrera <alvherre(at)2ndquadrant(dot)com>
wrote:

> Robert Haas wrote:
> > On Mon, Jun 6, 2016 at 11:50 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> > > Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> > >> On Mon, May 23, 2016 at 4:15 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> > >>> 2. Rewrite into LATERAL ROWS FROM (srf1(), srf2(), ...). This would
> > >>> have the same behavior as before if the SRFs all return the same
> number
> > >>> of rows, and otherwise would behave differently.
> > >
> > >> I thought the idea was to rewrite it as LATERAL ROWS FROM (srf1()),
> > >> LATERAL ROWS FROM (srf2()), ...
> > >
> > > No, because then you get the cross-product of multiple SRFs, not the
> > > run-in-lockstep behavior.
> >
> > Oh. I assumed that was the expected behavior. But, ah, what do I know?
>
> Lots, I assume -- but in this case, probably next to nothing, just like
> most of us, because what sane person or application would be really
> relying on the wacko historical behavior, in order to generate some
> collective knowledge? However, I think that it is possible that
> someone, somewhere has two SRFs-in-targetlist that return the same
> number of rows and that the current implementation works fine for them;
> if we redefine it to work differently, we would break their application
> silently, which seems a worse problem than breaking it noisily while
> providing an easy way forward (which is to move SRFs to the FROM list)
>
> My vote is to raise an error in the case of more than one SRF in
> targetlist.
>

​As long as someone is willing to put in the effort we can make a subset of
these multiple-SRFs-in-targetlist queries work without any change in the
tabular output, though the processing mechanism might change.​ Your vote
is essentially #1 up-thread which seems the most draconian. Assuming a
viable option 2.5 or 3 solution is presented would you vote against it
being committed? If so I'd like to understand why. I see #1 as basically
OK only if their are technical barriers we cannot overcome - including
performance.

Link to the definition of the various options Tom proposed:

https://www.postgresql.org/message-id/21076.1464034513%40sss.pgh.pa.us

David J.