Re: [oauth] Stabilize the libpq-oauth ABI (and allow alternative implementations?)

From: Jacob Champion <jacob(dot)champion(at)enterprisedb(dot)com>
To: Zsolt Parragi <zsolt(dot)parragi(at)percona(dot)com>
Cc: PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org>, Chao Li <li(dot)evan(dot)chao(at)gmail(dot)com>
Subject: Re: [oauth] Stabilize the libpq-oauth ABI (and allow alternative implementations?)
Date: 2026-03-03 22:08:21
Message-ID: CAOYmi+nCg5upBVOo_UCSjMfO=YMkZXcSEsgaADKXqerr5wahZQ@mail.gmail.com
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Fri, Feb 27, 2026 at 11:42 AM Jacob Champion
<jacob(dot)champion(at)enterprisedb(dot)com> wrote:
> 0001 should be ready to go. I meant to put 0002 up in its own thread
> at the beginning of the week but got sidetracked, so I'll go do that
> now.

Done at [1].

v5 incorporates that version of the patch in -0002 and makes some
minor test updates for better coverage in -0003.

> 0003 has to wait on Valgrind, and 0004-5 have to wait on 0002.

v5-0006 is my experiment in Valgrind (and also AddressSanitizer, and
also generic pointer poisoning). I think it's interesting, and it
seems clean enough for some subset to be committable for 19. The
weak-symbol support for ASan is the most suspect; I have no idea if it
links correctly across all platforms. It's probably better for me to
propose generic ASan support in utils/memdebug.h at some point in the
future, and drop it from this patch.

I don't plan to block the other patches on -0006, since -0004 is
blocking PGOAUTHCAFILE.

--Jacob

[1] https://postgr.es/m/CAOYmi%2B%3DMHD%2BWKD4rsTn0v8220mYfyLGhEc5EfhmtqrAb7SmC5g%40mail.gmail.com

Attachment Content-Type Size
since-v4.diff.txt text/plain 4.0 KB
v5-0001-oauth-Report-cleanup-errors-as-warnings-on-stderr.patch application/octet-stream 2.8 KB
v5-0002-libpq-Add-PQgetThreadLock-to-mirror-PQregisterThr.patch application/octet-stream 4.1 KB
v5-0003-libpq-Introduce-PQAUTHDATA_OAUTH_BEARER_TOKEN_V2.patch application/octet-stream 24.7 KB
v5-0004-libpq-oauth-Use-the-PGoauthBearerRequestV2-API.patch application/octet-stream 43.5 KB
v5-0005-libpq-oauth-Never-link-against-libpq-s-encoding-f.patch application/octet-stream 3.2 KB
v5-0006-WIP-test-out-poisoning.patch application/octet-stream 8.3 KB
v5-0007-WIP-Introduce-third-party-OAuth-flow-plugins.patch application/octet-stream 38.8 KB

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Masahiko Sawada 2026-03-03 22:23:45 Re: doc: add note that wal_level=logical doesn't set up logical replication in itself
Previous Message Ilia Evdokimov 2026-03-03 21:55:31 Re: Reduce planning time for large NOT IN lists containing NULL