Re: Pass ParseState as down to utility functions.

On Thu, Dec 12, 2024 at 10:08:04AM +0800, jian he wrote:
> On Thu, Dec 12, 2024 at 4:48 AM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> I'm not loving the idea of cons'ing up ParseStates in random places in
>> tablecmds.c. I think we ought to fix things so that the one made in
>> standard_ProcessUtility is passed down to all these places, replacing
>> ad-hoc queryString and queryEnv parameters.

> the main code change is within DefineDomain.

[ .. more comments from Jian .. ]

Yeah, I'm not much a fan of some of the changes of tablecmds.c, that
makes the whole stack more complicated than it should.

ProcessUtilitySlow() passing down a ParseState to DefineDomain() and
AlterType() is more consistent and a good idea IMO.

I think that the patch should be split a bit more. Even if different
areas of the code are touched, there's more than one idea of how to
create this information for the error position, so each piece could be
committed separately with changes showing up as diffs in the
regression tests.
--
Michael