On 19/8/2024 18:36, Tom Lane wrote:
> Andrei Lepikhov <lepihov@gmail.com> writes:
>> Thanks for pushing this!
>
>> But could you change this code a little bit?
>> I reported this issue a year ago. At that time, it was triggered by the
>> CustomScan node [1]. I haven't found the solution there [2]. Your code
>> looks like a good tradeoff, and if you slightly change the code (like in
>> the attachment), it allows CustomScan to survive such cases.
>
> This seems like it's making assumptions it shouldn't about what
> CustomScan does. If there's an argument for doing this, it should
> be added to the adjacent comments.
Hm, I got into this problem many times using CustomScan node. Do you
have some objections to not allow CustomScan node have a RECORD Var in
the target list? For example, once I got this bug designing CustomScan
which gathered lightweight statistics on-the-fly under a Sort and
GROUP-BY nodes. I didn't change any tuple and had the same target list
as the child node. Why we should analyse target list and don't use
CustomScan if it contains Var of specific type?
>
> (Also, what's with the random change in contrib/Makefile?)
Oops, it's a waste code, pardon.
--
regards, Andrei Lepikhov
