2025-11-07
Shibboleth Developer's Meeting, 2025-11-07
Call Administrivia
09:00 Central US / 10:00 Eastern US / 15:00 UK / 17:00 FI
Calls are normally the 1st and 3rd Fridays of each month. Next call would be Friday 2025-11-21. Any reason to deviate from this?
60 to 90 minute call window.
Call Details
This week's call will use the Zoom system at GU, see ZoomGU for access info.
AGENDA
Please review any IdP 5.2.0 backlog
SP POST preservation – denial of service mitigations?
Board update
Attendees:
Brent
https://shibboleth.atlassian.net/browse/JSATTR-6
A review of some base class behavior (fail-fast, error options, etc) and minor items pending
Otherwise mostly down to just testing and tweaking
https://shibboleth.atlassian.net/browse/OSJ-434
Review shows that (I think) reporter’s observations are correct and we have (at least) 2 obvious errors
Easy fixes, but how much do we want to go to “support” Java modules?
Daniel
Henri
First iteration of API vs IMPL division for the current fed-plugin code
Hard to avoid fed-impl being dependent on OP-impl
Explicit registration improvements
Now works better with remote trust chain resolution
Trust Chain + Trust Mark data is stored to the client information record
Can then be exploited during the OAuth2/OIDC sequences
Improved flow tests (e.g. content-type and JWT type headers are now up to date)
Log fine-tuning according to feedback from the eduGAIN pilot deployers
Some are not straight-forward (e.g. warnings from metadata cache)
Ian
https://shibboleth.atlassian.net/wiki/spaces/DEV/pages/4629659652 :
testingwas deleted.Seems to be working: “I completely forgot we switched the SP plugins” (Scott).
Exception: commits pushed to Codeberg and mirrored to
git.shibboleth.netare not being mailed to thecommits@mailing list. I don’t know why and it’s hard to debug. I have tried a couple of tweaks to the mailer without success. How much does this matter?Some progress with this, maybe.
Are we at the point where this isn’t an experiment any more?
Next steps other than that declaration? (Next repos to formally move):
xmlsectool(Ian)java-idp-translations(translators are special cased)Things we’ve worked on in the last couple of weeks, per the current mirroring script:
cpp-sp,java-identity-provider,java-idp-plugin-oidc-op-oidfed,java-plugin-webauthn,java-oidc-common,java-parent-projectMore disruptive, but that can be seen as a good thing as it’s going to flush out any issues.
Things that are current but haven’t been worked on in the last couple of weeks:
Somewhat disruptive, but not much feedback available.
Things that are not current (e.g.,
java-support)Not disruptive at all, but correspondingly low benefit. Need to be done eventually.
Could even set these on Codeberg as not belonging to any teams, preserving the “locked” behaviour some of them have in Gitolite already.
I’d be tempted to save effort by not setting up Codeberg to
git.shibboleth.netmirroring for these, make them static, and ultimately remove them from our server.
John
https://shibboleth.atlassian.net/browse/CPPSP-26
SP4 alpha 1 is building on all supported platforms.
Currently: linting/scrubbing SPEC file
Next: remove unneeded built-time packages from builder images
Later: smoke test installs including SxS w/SP3
Marvin
Phil
WebAuthn:
2 bug fixes, done.
1 feature request, need to think about it
OIDC RP:
TokenConsumer flow now performing back-channel Token requests
And failing, I will move my Mock server over from the RP-Proxy for testing
Next would be to plugin in the id_token validation logic
Then, UserInfo lookup and back out as a response.
Trying to figure out how best to manage state, and what needs to be saved off.
Rod
Playing with SP4 hub and agents (IIS and Apache on Windows).
It's brain bending juggling 4 different machine (Hub, IdP, and 2 agents) but it all just works.
But it has been a chance to use the jetty plugin for real (in two different places)
All my issues have been PEBKAC or in my code.
Next up WiX (and much procrastination)
Jetty-base plugin
Being tweaked.
I did many of the the changes proposed 3 weeks ago including separate base, IdP and SP modules
5.2 backlog
Will try to get IDP-2378 at least partially completed. The rest can be closed, pushed out or punted to the jetty plugin
Potential new work item (to auto-set IDP_BASE_URL) outlined in JJETTY-24
Scott
SP documentation ongoing, getting into the meat of the Hub, which is much harder to document due to Spring, etc.
Inheritance material is incomprehensible but had to park for now
Ported up rest of legacy access control features form V3
Had to hack in some kind of ISO 8601 parsing to the best I could manage, stuck to UTC for full time stamps, which was a change from V3
Sacrificed compatibility with the more complex legacy cases to consolidate the old messy syntax into a single “schema” and “XML” plugin type for all the supported rules
https://shibboleth.atlassian.net/browse/JSHIBD-16
Fully indirected the basic auth “username” from the underlying Agents in the Hub to support N:1 service account/agent relationship
Adjusted how agent ID is handled in Agent by moving it to a [remoting] setting, emptying out [global] section
Continued fixes to defaults based on Rod’s testing
Remaining major work items per blog:
Alternative Session Cache options, including the one we may end up using later as a default.
Cleaning up and finalizing some of the error handling features in the Agent
Audit logging in the Hub
OIDC support (probably absent the Federation support initially)
Logout – I’d like to at least get a local logout feature working for alpha
Documentation, documentation, and documentation
Tom
Jenkins and Java maintenance
Working on testing Jetty and SP plugins
Other