Re: Optimize common expressions in projection evaluation

On Sun, Dec 4, 2022 at 9:37 PM Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com>
wrote:

>
> po 5. 12. 2022 v 5:28 odesílatel Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> napsal:
>
>> Peifeng Qiu <pgsql(at)qiupf(dot)dev> writes:
>> >> the need for this code seems not that great. But as to the code
>> itself I'm unable to properly judge.
>>
>> I mention this because trying to reverse-engineer this situation
>> in execExpr seems seriously ugly and inefficient, even assuming
>> you can make it non-buggy. The right solution has to involve never
>> expanding foo().* into duplicate function calls in the first place,
>> which is the way it used to be. Maybe if you dug around in those
>> twenty-year-old changes you could get some inspiration.
>>
>> I tend to agree with David that LATERAL offers a good-enough
>> solution in most cases ... but it is annoying that we accept
>> this syntax and then pessimize it.
>>
>
> I agree, so there is a perfect solution like don't use .*, but on second
> hand any supported syntax should be optimized well or we should raise some
> warning about negative performance impact.
>
>
Yeah, I phrased that poorly. I agree that having this problem solved would
be beneficial to the project. But, and this is probably a bit unfair to
the OP, if Tom decided to implement LATERAL after years of hearing
complaints I doubted this patch was going to be acceptable. The
benefit/cost ratio of fixing this in code just doesn't seem to be high
enough to try at this point. But I'd be happy to be proven wrong. And I
readily admit both a lack of knowledge, and that as time has passed maybe
we've improved the foundations enough to implement something here.

Otherwise, we can maybe improve the documentation. There isn't any way
(that the project would accept anyway) to actually warn the user at
runtime. If we go that route we should probably just disallow the syntax
outright.

David J.