Re: some more include removal from headers

Hello,

On 2026-Mar-13, Andres Freund wrote:

> On 2026-03-13 20:32:53 +0100, Álvaro Herrera wrote:

> > Not sure about stringinfo.h, which doesn't change very often and doesn't
> > include anything else.
>
> It seemed worth it because it's not used by plannodes.h, it's a leftover from
> before a05dc4d7fd5.

Oh, interesting. It's fine with me of course.

> > I'm doubtful about the bitmapset.h removal gaining
> > us much either (because the really nasty one below, nodetags.h, is
> > unavoidable anyway.)
>
> Yea, I was very much on the fence with that one.

Let's leave it out for now. We can always think about it later.

> Agreed, we can tackle that one separately. There's multiple paths for it too,
> e.g. a separate header for the instr_time type or moving towards not not
> needing to include instrument[_node].h.

Right ...

> How would you like to work towards committing these? I'm am more than happy
> for you to commit everything, but I could also help.

I can try to find time over the weekend to push it all, assuming a
single commit is okay. If we want more than one then I'm afraid it'd
have to wait until sometime next week. But I don't think that is
justified.

--
Álvaro Herrera PostgreSQL Developer — https://www.EnterpriseDB.com/
"I love the Postgres community. It's all about doing things _properly_. :-)"
(David Garamond)