Re: Allow file inclusion in pg_hba and pg_ident files

Hi,

On Wed, Mar 23, 2022 at 11:03:46AM +0900, Michael Paquier wrote:
>
> Pushing forward with 0001 by the end of the CF is the part that has no
> controversy IMO, and I have no objections to it. Now, after looking
> at this part, I found a few things, as of:
> - HbaToken, the set of elements in the lists of TokenizedAuthLine, is
> a weird to use as this layer gets used by both pg_hba.conf and
> pg_indent.conf before transforming them into each HbaLine and
> IdentLine. While making this part of the internals exposed, I think
> that we'd better rename that to AuthToken at least. This impacts the
> names of some routines internal to hba.c to copy and create
> AuthTokens.

Yeah, I thought about it but didn't rename it given your concerns about git
history. I'm fine either way.

> - s/gethba_options/get_hba_options/, to be consistent with
> fill_hba_view() and other things.
> - The comment at the top of tokenize_auth_file() needed a refresh.
>
> That's mostly cosmetic, and the rest of the code moved is identical.
> So at the end this part looks rather commitable to me.

Looks good to me, thanks.

> I have not been able to test 0002 in details, but it looks rather
> rather sane to me at quick glance, and it is simple. The argument
> about more TAP tests applies to it, though, even if there is one SQL
> test to check the function execution. It is probably better to not
> consider 0003 and 0004 for this CF.

No objection to moving 0003 and 0004 to the next commitfest.