Re: WAL record CRC calculated incorrectly because of underlying buffer modification

On Thu, 2026-10-29 at 22:51 -0400, shihao zhong wrote:
> I agree with that, but I think there are still a decent number of
> users
> who are using hash indexes for different reasons.

...

> Looks like this is an interesting problem. I do agree that if there
> is
> any potential optimization for btree, we should take it all. If
> nobody
> volunteers, I can create a different thread and create a POC for it
> after 19 GAed.

Yes, that makes sense to me. If you think there's benefit to be had,
please go ahead in a new thread.

>
> I am not sure. I know some types' = has special logic, where it may
> not
> mean they are the same.
>
> For example, 'ABC' = 'abc' is true
> with a case-insensitive collation. But md5() gives different results.

Right, there are a few details to work out. But I do think there's room
to solve this in the optimizer rather than add more special cases to
index DDL and the AMs.

We could start by saying it only works if f() is part of a hash
opclass, or something.

Regards,
Jeff Davis