Shibboleth Developer's Meeting, 2022-01-07

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 2022-01-21. Any reason to deviate from this?

60 to 90 minute call window.

This week's call will use the Zoom system at GU, see ZoomGU for access info.

Jetty 10 (Rod/Scott)
https://spring.io/blog/2022/01/05/spring-framework-cve-2021-22060-has-been-published
Ian Young Can Jenkins use parallel builds for -multi jobs? Only one executor per worker, of course.
Rod Widdowson IDP-1793 - Getting issue details... STATUS

Out for last 2.5 weeks, nothing to report
Will look at OSJ-346. Perhaps there isn’t normative guidance in spec on IssueInstant in Assertions.

JOIDC-65 - Getting issue details... STATUS
- The main reason for the 3.0.3 patch
JCOMOIDC-28 - Getting issue details... STATUS
- Implemented metadata policy merging, as specified by OIDCfed draft - in our case, we need it for merging policies defined in the profile configuration (dyn.reg.) and access token
JOIDC-61 - Getting issue details... STATUS
- Pushed the initial (still a bit incomplete) implementation

JPAR-186 - Getting issue details... STATUS
- java-support, spring-extensions, java-opensaml relatively easy
- IdP harder, requires Spring Web Flow, leading to:
IDP-1892 - Getting issue details... STATUS
- We have a ticket https://jira.spring.io/browse/SWF-1750 upstream asking whether this will be ported (no response yet, assuming this isn’t going to happen).
- I have a fork at https://github.com/iay/spring-webflow/tree/dev/IDP-1892 which is 90% of the way (remains: POM generation) to building the current SWF under Java 17 and (required for that) Gradle 7.0. This is (probably) the hard part, just using the EE9 APIs should be straightforward at that point… unless there are other dependencies I don’t know about yet.
Aim here is not to build a production stack, but to find out what would block us from doing so. I’d hope most of the actual code I’m writing would be useable long term, but it’s not the primary goal.
Random asides:
- This is the first thing I’ve had to do with Gradle. I find myself hating it less than Maven just now. This may just be the optimism of unfamiliarity, but its basic model is better in several ways.
- SWF was very out of touch with its build system. If they do update it themselves, they might well not go the same way I did to fix the issues.

Mostly maintenance chores: spinning up a new Rocky 8 test instance and dealing with Docker/Podman silliness, SELinux silliness, AWS silliness, etc to re-test cpp-linbuild; updated Amazon Linux 2 image. Still need to re-constitute my rudimentary Jenkins instance on which to hack, and then recall where I was with Fargate-based builders before the holidays.

JCOMOIDC-30 - Getting issue details... STATUS
- Message encoders to support query string serialization (GET) and form serialization (POST) for authentication requests.
- A factory that returns an encoder based on which type is configured in the profile configuration.
JCOMOIDC-35 - Getting issue details... STATUS
- Mutable objects to help build an OIDC authn (OAuth authz) request.
Improvements to the RP flow up to making authn/authz requests via the controller.

Enforcer
- Now (using the 3.1-SNAPSHOT) running with up to date and XMLSec friendly data
- This includes the ~/.m2/respository thing for the idp Nightlight (but see JMVN-13 - Getting issue details... STATUS for why this is fragile)
- Need to do a release of 3.1 for the enforcer and 1.0 of the data soon (preferably before the next release of a distributable)
Logback - prepared the windows jetty-base to ship logback latest logback.

JOIDC-11 - Getting issue details... STATUS and dependent work
- Extending client authentication done, working through token flow changes
- Will need additional changes to introspection/revocation flows to add access control (use existing AccessControl layer I hope?)
- Leaning to relying on Attribute Resolver somehow to produce scope and audience claims for tokens