Re: [PoC] Federated Authn/z with OAUTHBEARER
Daniel Gustafsson <daniel@yesql.se>
Commits
GET /api/v1/messages/:b64id/commits
the thread's linked commits as JSON, with link sources.
API reference →
-
meson: Fix install-quiet after clean
- a9ffb35274fb 18.0 landed
- 4ae03be54734 19 (unreleased) landed
-
oauth: Run Autoconf tests with correct compiler flags
- 3d23f68c5529 18.0 landed
- 990571a08b66 19 (unreleased) landed
-
Link libpq with libdl if the platform needs that.
- 4df477153a6b 19 (unreleased) landed
- 7bd752c1fb8e 18.0 landed
-
Doc: correct spelling of meson switch.
- 3faac9d14063 16.9 landed
- 766d2e673342 17.5 landed
- ac557793d478 18.0 landed
-
oauth: Correct SSL dependency for libpq-oauth.a
- 3db68212a393 18.0 landed
-
oauth: Fix Autoconf build on macOS
- 4ea1254f35b2 18.0 cited
-
oauth: Move the builtin flow into a separate module
- b0635bfda053 18.0 landed
-
Remove a stray "pgrminclude" annotation
- 764d501d24ba 18.0 cited
-
oauth: Simplify copy of PGoauthBearerRequest
- 1cf4c56480f8 18.0 landed
-
oauth: Improve validator docs on interruptibility
- 873c0fd67872 18.0 landed
-
oauth: Disallow synchronous DNS in libcurl
- d7e40845f923 18.0 landed
-
oauth: Fix postcondition for set_timer on macOS
- 434dbf6907ec 18.0 landed
-
oauth: Use IPv4-only issuer in oauth_validator tests
- 8d9d5843b55f 18.0 landed
-
Work around OAuth/EVFILT_TIMER quirk on NetBSD.
- c301a0a74a8a 18.0 landed
-
oauth: Fix incorrect const markers in struct
- 03366b61dfe5 18.0 landed
-
Add missing entry to oauth_validator test .gitignore
- 2c53dec7f440 18.0 landed
-
cirrus: Temporarily fix libcurl link error
- 9d9a71002a1c 18.0 landed
-
Add support for OAUTHBEARER SASL mechanism
- b3f0be788afc 18.0 landed
-
libpq: Handle asynchronous actions during SASL
- a99a32e43ed7 18.0 landed
-
require_auth: prepare for multiple SASL mechanisms
- f8d8581ed882 18.0 landed
-
Move PG_MAX_AUTH_TOKEN_LENGTH to libpq/auth.h
- e21d6f297158 18.0 landed
-
Make SASL max message length configurable
- 6d16f9debae0 18.0 landed
-
jsonapi: fully initialize dummy lexer
- 41b023946dfd 18.0 landed
-
common/jsonapi: support libpq as a client
- 0785d1b8b2fa 18.0 landed
-
Remove fe_memutils from libpgcommon_shlib
- f1976df5eaf2 18.0 landed
-
Revert ECPG's use of pnstrdup()
- f0096ef13be2 13.17 landed
- 3557185538fe 14.14 landed
- 2de129b356bf 15.9 landed
- ee2997c678d8 16.5 landed
- e9e05c655069 17.0 landed
- 5388216f6adc 18.0 landed
-
Explicitly require password for SCRAM exchange
- adcdb2c8dda4 17.0 landed
-
Refactor SASL exchange to return tri-state status
- 24178e235ea5 17.0 landed
> On 20 Dec 2024, at 02:00, Jacob Champion <jacob.champion@enterprisedb.com> wrote:
> v40 also contains:
A few small comments on v40 rather than saving up for a longer email:
+ ereport(LOG, errmsg("Internal error in OAuth validator module"));
Tiny nitpick, the errmsg() should start with lowercase 'i'.
+ libpq_append_conn_error(conn, "no custom OAuth flows are available, and libpq was not built using --with-libcurl");
Since we at some point will move away from autoconf I think we should avoid
such implementation details in error messages. How about "was not built with
libcurl support"?
+ * Find the start of the .well-known prefix. IETF rules state this must be
+ * at the beginning of the path component, but OIDC defined it at the end
+ * instead, so we have to search for it anywhere.
I was looking for a reference for OIDC defining the WK prefix placement but I
could only find it deferring to RFC5785 like how RFC8414 does. Can you inject
a document reference for this?
+ if (strcmp(conn->oauth_issuer_id, discovery_issuer) != 0)
Shouldn't the scheme component really be compared case-insensitive, or has it
been normalized at this point? Not sure how much it matters in practice but if
not perhaps we should add a TODO marker there?
Support for oauth seems to be missing from pg_hba_file_rules() which should be
added in hbafuncs.c:get_hba_options().
--
Daniel Gustafsson