MIME-Version: 1.0
References: 
 <CAOYmi+mrGg+n_X2MOLgeWcj3v_M00gR8uz_D7mM8z=dX1JYVbg@mail.gmail.com>
 <3720B2E1-0B96-4063-8D63-B5AE6AFEA159@gmail.com>
 <CAOYmi+mwzS4xfgomL1Lte+W1hRk+zahhZLnRMDJ8nYxoHUWt_w@mail.gmail.com>
 <CAOYmi+mEU_q9sr1PMmE-4rLwFN=OjyndDwFZvpsMU3RNJLrM9g@mail.gmail.com>
 <CAN4CZFPwdH_GMrappThju=t4m5n95Dse6ecTNwN=h4GpZLOMng@mail.gmail.com>
 <CAOYmi+m-hphbx=CRfGSdcrHi-jL3OrZGJMTAPOYp=OK253hBTA@mail.gmail.com>
 <CAOYmi+nnYr_Oe=KH2ZW6Ld88eZWv5P-s5ZtWOEh2KsPtMqOzbg@mail.gmail.com>
 <CAOYmi+nCg5upBVOo_UCSjMfO=YMkZXcSEsgaADKXqerr5wahZQ@mail.gmail.com>
 <CAOYmi+nRmt+K+ZVNztKFv=LuasDCCSsXLihZB6BEkVeroW8EvA@mail.gmail.com>
 <CAN4CZFOMv=0Y1PZ8uw3xgJmEqt4D1dr1-yOq9jjZYTARZp7H9Q@mail.gmail.com>
 <CAOYmi+mLaohk3FLbH9fKGmzN7yFGHs2Zefdyp31wbL1130puiA@mail.gmail.com>
 <CAOYmi+=Pr7AAdkcKXyLw3ycxcrjGKsOV2CTYVV2PKYQw9ecG0Q@mail.gmail.com>
 <be219c8a903f9c8a190c503cd5a6d084a56fd9a5.camel@gmail.com>
 <CAOYmi+=cd0SdEm4Pw3RE4eD=rbXex9Rkc3DsNiaDAqQti9taQQ@mail.gmail.com>
 <821576A8-5958-401D-B8D1-7E4E30F5A40F@gmail.com>
In-Reply-To: <821576A8-5958-401D-B8D1-7E4E30F5A40F@gmail.com>
From: Jacob Champion <jacob.champion@enterprisedb.com>
Date: Tue, 31 Mar 2026 11:50:02 -0700
Message-ID: 
 <CAOYmi+=78NhN+6K-p15Q6vJWYCvsuQtB5swM7g6wRn+gBWU0eA@mail.gmail.com>
Subject: Re: [oauth] Stabilize the libpq-oauth ABI (and allow alternative
 implementations?)
To: Chao Li <li.evan.chao@gmail.com>
Cc: "Jonathan Gonzalez V." <jonathan.abdiel@gmail.com>,
 Zsolt Parragi <zsolt.parragi@percona.com>,
	PostgreSQL Hackers <pgsql-hackers@postgresql.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Archived-At: 
 <https://www.postgresql.org/message-id/CAOYmi%2B%3D78NhN%2B6K-p15Q6vJWYCvsuQtB5swM7g6wRn%2BgBWU0eA%40mail.gmail.com>
Precedence: bulk

On Mon, Mar 16, 2026 at 12:33=E2=80=AFAM Chao Li <li.evan.chao@gmail.com> w=
rote:
> I think this check relies on that when poisoning, request->error should b=
e NULL. So, does it make sense to Assert(request->error=3D=3DNULL) in the p=
oison branch?

Yes, that's a good idea. Done in the committed version.

> 2 - 0002 Overall LGTM. A small comment is that, now use_builtin_flow() be=
comes a bit misleading because it may turn to non-builtin when libpq_oauth_=
init is not provided by the lib. So, maybe rename the function to something=
 like try_builtin_flow()?

I think you're correct that the naming here has not caught up to where
we are, but try_builtin_flow() doesn't get us much closer in my
opinion, and I wasn't able to find something I really liked. This will
need a refactor when plugins arrive, so I'm deferring for now.

Thanks!
--Jacob