Horton

Thread

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Declare load_hosts() as returning HostsFileLoadResult.

  1. Re: Serverside SNI support in libpq

    Tom Lane <tgl@sss.pgh.pa.us> — 2026-05-04T19:22:37Z

    In preparation for our annual pgindent update, I checked what happens
when I install the buildfarm's version of typedefs.list, and I found
that the typedef HostsFileLoadResult (from 4f433025f) gets
misformatted because it's not in the buildfarm's list.  That's because
the buildfarm mechanism only captures typedefs that are used to
declare some object (variable, function, field) and this one isn't.

It seems quite odd to me that load_host(), which in fact returns
HostsFileLoadResult codes, is declared to return int.  That seems
to have been done because HostsFileLoadResult wasn't declared in
the same header, but there is no visible reason why it shouldn't be.
Any objection to the attached fixup?

As a side matter, "load_host" seems like a remarkably generic name
that conveys little about what it actually does, and to the extent
that it does convey anything the implication is wrong: it returns
(potentially) info about multiple hosts not just one.  Can't we do
better?

			regards, tom lane

  2. Re: Serverside SNI support in libpq

    Tom Lane <tgl@sss.pgh.pa.us> — 2026-05-04T19:27:20Z

    I wrote:
> As a side matter, "load_host" seems like a remarkably generic name
> that conveys little about what it actually does, and to the extent
> that it does convey anything the implication is wrong: it returns
> (potentially) info about multiple hosts not just one.  Can't we do
> better?

Sigh ... brain fade there, of course the function is load_hosts
not load_host.  It's still too generic IMO, but at least the
pluralization is right.

			regards, tom lane

  3. Re: Serverside SNI support in libpq

    Daniel Gustafsson <daniel@yesql.se> — 2026-05-04T20:31:36Z

    > On 4 May 2026, at 21:22, Tom Lane <tgl@sss.pgh.pa.us> wrote:

> It seems quite odd to me that load_host(), which in fact returns
> HostsFileLoadResult codes, is declared to return int.  That seems
> to have been done because HostsFileLoadResult wasn't declared in
> the same header, but there is no visible reason why it shouldn't be.
> Any objection to the attached fixup?

At some point during the development of the patch there was a reason (which I
cannot remember right now) for the declaration being in hba.h, but I clearly
missed moving it when that no longrer applied.  No objections to the patch,
thanks!

> As a side matter, "load_host" seems like a remarkably generic name
> that conveys little about what it actually does, and to the extent
> that it does convey anything the implication is wrong: it returns
> (potentially) info about multiple hosts not just one.  Can't we do
> better?

It's following the naming convention of load_hba() which reads pg_hba.conf, and
load_ident() which reads pg_ident.conf - thus load_hosts() for the function
that reads pg_hosts.conf.  Perhaps load_pg_hosts_conf() or load_hosts_config()
would convey more meaning?

--
Daniel Gustafsson

  4. Re: Serverside SNI support in libpq

    Tom Lane <tgl@sss.pgh.pa.us> — 2026-05-04T22:00:02Z

    Daniel Gustafsson <daniel@yesql.se> writes:
>> On 4 May 2026, at 21:22, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> As a side matter, "load_host" seems like a remarkably generic name
>> that conveys little about what it actually does,

> It's following the naming convention of load_hba() which reads pg_hba.conf, and
> load_ident() which reads pg_ident.conf - thus load_hosts() for the function
> that reads pg_hosts.conf.  Perhaps load_pg_hosts_conf() or load_hosts_config()
> would convey more meaning?

Hmm, okay.  I'd prefer a more specific name, but it wouldn't make
much sense unless we also rename those two.  That's probably more
code churn than is justified.

I'll push the thing for moving/using the typedef, but leave the
function name alone.

			regards, tom lane

  5. Re: Serverside SNI support in libpq

    Daniel Gustafsson <daniel@yesql.se> — 2026-05-05T07:40:19Z

    > On 5 May 2026, at 00:00, Tom Lane <tgl@sss.pgh.pa.us> wrote:

> I'll push the thing for moving/using the typedef, but leave the
> function name alone.

Thanks!

--
Daniel Gustafsson