Re: using extended statistics to improve join estimates

On Sun, Apr 28, 2024 at 10:07:01AM +0800, Andy Fan wrote:
> 's/estimiatedcluases/estimatedclauses/' typo error in the
> commit message is not fixed since I have to regenerate all the commits

Maybe you know this, but some of these patches need to be squashed.
Regenerating the patches to address feedback is the usual process.
When they're not squished, it makes it hard to review the content of the
patches.

For example:
[PATCH v1 18/22] Fix error "unexpected system attribute" when join with system attr
..adds .sql regression tests, but the expected .out isn't updated until
[PATCH v1 19/22] Fix the incorrect comment on extended stats.

That fixes an elog() in Tomas' original commit, so it should probably be
002 or 003. It might make sense to keep the first commit separate for
now, since it's nice to keep Tomas' original patch "pristine" to make
more apparent the changes you're proposing.

Another:
[PATCH v1 20/22] Add fastpath when combine the 2 MCV like eqjoinsel_inner.
..doesn't compile without
[PATCH v1 21/22] When mcv->ndimensions == list_length(clauses), handle it same as

Your 022 patch fixes a typo in your 002 patch, which means that first
one reads a patch with a typo, and then later, a 10 line long patch
reflowing the comment with a typo fixed.

A good guideline is that each patch should be self-contained, compiling
and passing tests. Which is more difficult with a long stack of
patches.

--
Justin