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

On January 18, 2017 3:59:00 PM PST, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
>On Wed, Jan 18, 2017 at 6:19 PM, Andres Freund <andres(at)anarazel(dot)de>
>wrote:
>>> SELECT x, CASE WHEN x > 0 THEN generate_series(1, 5) ELSE 0 END
>FROM tab;
>>>
>>> It might seem that this should produce five repetitions of input
>rows
>>> that have x > 0, and a single repetition of those that do not; but
>>> actually it will produce five repetitions of every input row. This
>is
>>> because generate_series() is run first, and then the CASE
>expression is
>>> applied to its result rows. The behavior is thus comparable to
>>>
>>> SELECT x, CASE WHEN x > 0 THEN g ELSE 0 END
>>> FROM tab, LATERAL generate_series(1,5) AS g;
>>>
>>> It would be exactly the same, except that in this specific
>example, the
>>> planner could choose to put g on the outside of the nestloop join,
>since
>>> g has no actual lateral dependency on tab. That would result in a
>>> different output row order. Set-returning functions in the select
>list
>>> are always evaluated as though they are on the inside of a
>nestloop join
>>> with the rest of the FROM clause, so that the function(s) are run
>to
>>> completion before the next row from the FROM clause is considered.
>>>
>>> So is this too ugly to live, or shall we put up with it?
>>
>> I'm very tentatively in favor of living with it.
>
>So, one of the big reasons I use CASE is to avoid evaluating
>expressions in cases where they might throw an ERROR. Like, you know:
>
>CASE WHEN d != 0 THEN n / d ELSE NULL END
>
>I guess it's not the end of the world if that only works for
>non-set-returning functions, but it's something to think about.

That's already not reliable in a bunch of cases, particularly evaluation during planning... Not saying that's good, but it is.

Andres
--
Sent from my Android device with K-9 Mail. Please excuse my brevity.