Re: Implement targetlist SRFs using ROWS FROM() (was Changed SRF in targetlist handling)

Hi,

On 2017-01-18 14:24:12 -0500, Tom Lane wrote:
> Andres Freund <andres(at)anarazel(dot)de> writes:
> > On 2017-01-18 08:43:24 -0500, Tom Lane wrote:
> >> ... except for one thing. The more I look at it,
> >> the more disturbed I am by the behavioral change shown in rangefuncs.out
> >> --- that's the SRF-in-one-arm-of-CASE issue.
>
> > I'm fine with leaving it as is in the patch, but I'm also fine with
> > changing things to ERROR. Personally I don't think it matters much, and
> > we can whack it back and forth as we want later. Thus I'm inclined to
> > commit it without erroring out; since presumably we'll take some time
> > deciding on what exactly we want to prohibit.
>
> I agree. If we do decide to throw an error, it would best be done in
> parse analysis, and thus would be practically independent of this patch
> anyway.

Cool, agreed then.

> >> * This bit in ExecProjectSRF was no good:
> >> + else if (IsA(gstate->arg, FuncExprState) &&
> >> + ((FuncExpr *) gstate->arg->expr)->funcretset)
>
> > Argh. That should have been FunExprState->func->fn_retset.
>
> Nope; that was my first thought as well, but fn_retset isn't valid if
> init_fcache hasn't been run yet, which it won't have been the first time
> through.

Righty-O :(

> So I think we can push this patch now and get on with the downstream
> patches. Do you want to do the honors, or shall I?

Whatever you prefer - either way, I'll go on to rebasing the cleanup
patch afterwards (whose existance should probably be mentioned in the
commit message).

Greetings,

Andres Freund