From: Tom Lane <tgl@sss.pgh.pa.us>
To: Nathan Bossart <nathandbossart@gmail.com>
cc: Michael Paquier <michael@paquier.xyz>, pgsql-hackers@postgresql.org
Subject: Re: pg_parameter_aclcheck() and trusted extensions
In-reply-to: <20220707184001.GA2255643@nathanxps13>
References: <20220706224727.GA2158260@nathanxps13>
 <YsYxEtX8zOYKf3Ru@paquier.xyz> <2995921.1657212060@sss.pgh.pa.us>
 <20220707184001.GA2255643@nathanxps13>
Comments: In-reply-to Nathan Bossart <nathandbossart@gmail.com>
	message dated "Thu, 07 Jul 2022 11:40:01 -0700"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <3032102.1657220409.1@sss.pgh.pa.us>
Date: Thu, 07 Jul 2022 15:00:09 -0400
Message-ID: <3032103.1657220409@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/3032103.1657220409%40sss.pgh.pa.us>
Precedence: bulk

Nathan Bossart <nathandbossart@gmail.com> writes:
> On Thu, Jul 07, 2022 at 12:41:00PM -0400, Tom Lane wrote:
>> Can we get away with doing these things in beta3?  We could avoid
>> breaking (2) in the v15 branch by making set_config_option into
>> a wrapper around set_config_option_ext, or something like that;
>> but the problem with struct config_generic seems inescapable.

> I personally lean more towards the compatibility break than reverting the
> feature.  There are still a couple of months before 15.0, and I suspect it
> won't be too difficult to fix any extensions that break because of this.

I checked http://codesearch.debian.net and found only a couple of
extensions that #include guc_tables.h at all, so I'm satisfied
that the struct config_generic ABI issue is tolerable.  Recompiling
after beta3 would be enough to fix any problem there, and it's
hard to believe that anyone is trying to ship production-ready
v15 extensions already.

The aspect that is a bit more debatable is whether to trouble with
a set_config_option() wrapper to avoid the API break in v15.
I think we'd still be making people deal with an API break in v16,
so making them do it this year rather than next doesn't seem like
a big deal ... but maybe someone wants to argue it's too late
for API breaks in v15?

			regards, tom lane