GeoTools / GeoServer PMC meeting - 2026-06-02
Attending
-
Torben Barsballe
-
Kevin Smith
-
Peter Smythe
-
Pierre Mauduit
-
Jody Garnett
Actions from prior meetings:
N/A
Agenda
-
GeoServer 3.0 RC Feedback & Release
-
GeoTools XMLUtils
-
PR Review / Catch-up
-
Improve GitHub workflows stability/reliability
Actions
- Various: Review MRs listed under 3.
GeoServer 3.0 RC Feedback & Release
See Feedback thread here. Other feedback on the users list
How much testing & feedback are we expecting / do we want before moving forward?
Haven’t got a lot, but it’s generally been positive or has been addressed by a MR
For OIDC - advised to use nightlies, so testers get the applicable fixes.
-
What does this need for extension? GSIP-239
-
We have a few public responses - yay!
-
And a migration guide was being written by Alessio, wish to check in on that.
-
action: Alessio migration guide for OIDC
-
GeoServer 3.0 Readiness:
-
Docs:
-
Screen snaps in core / extensions - thanks for recent PRs
-
Some work to do still
-
Spreadsheet is a little out-of-date, should be updated based on recent PR activity
-
Who is available?
-
Jody is available to work on this “later in the week”, anyone else?
-
this is a new stable, so new branches and build jobs needed
-
Some release script work will be need due to change from 2.x-SNAPSHOT to 3.x.x-SNAPSHOT
- initial proposal was last-moment-rejected
-
-
More volunteers welcome
What happened to GeoServer 3.0 GSIP-226:
-
To avoid “code freeze” andrea asked that we break out a 3.0.0.x branch between release-build and release publish.
-
Insight: This will actually be good because any changes to release script will occur on main while the 3.0.0.x (and 3.0.0 tag) can remain unchanged.
-
Action: Make new release scripts for GeoServer 3, rather than try and maintain one script that works with 2.x and 3.x (see release breakage due to removal of docs/en/requirements.txt needed for release script)
-
So main branch will use build_release3, and then publish_release3
-
So 2.28.x uses build_release.sh, and then publish_release.sh
-
Note jenkins_release.sh looks at version numbers, so it can take care of calling the correct script above
-
aside: We have committed the release scripts to avoid scripts in Jenkins config being lost. Jenkinfile is another “workflow based” approach available to release from a tag https://www.jenkins.io/doc/book/pipeline/jenkinsfile/ similar to github actions
-
GeoTools XMLUtils
-
Jody and Pierre working on XMLUtils Integration
-
Now working and merging! Big improvement
-
Wasn’t able to reproduce prior app-schema failures (from GitHub workflows) locally, build is green now
-
-
See user manual for guidance
- There are new entity resolvers in GeoTools …
InternalEntityResolver for test cases (that can access jar:file and file limited to target/test-classes and target/classes (so obviously only for tests and local IDE development eh?)
- There are new entity resolvers in GeoTools …
PR Review / Catch-up
-
Replace mkdir call to more recent java.nio.files API geoserver/pull/9501 awaiting review
- action: Jody will review after meeting
-
GWC seeder thread pool:
-
https://github.com/GeoWebCache/geowebcache/pull/1535 and https://github.com/geoserver/geoserver/pull/9529
-
action: pmauduit will review after meeting
-
-
Deterministic XML output (order)
-
https://github.com/geotools/geotools/pull/5682 and https://github.com/geoserver/geoserver/pull/9537
- SLD file output changes based “random” due to HashMap use
-
Discussion about preserving order LinkedHashMap vs sorting TreeMap.
-
preserving order may be nice for folks that written by hand?
-
It will also keep the files that exist from changing?
-
-
action: Peter will update to LinkedHashMap, jody review when ready
-
-
XMLUtils default settings change:
-
action: pmauduit will review when jody takes out of draft
Improve GitHub workflows stability/reliability [GEOS-12127]
Various PRs have a tendency to fail CI tests the first time around and succeed when rerun (without code changes)
Discussion on Discourse - removal of webdav extension (and others)
-
and more…
-
Note, these are AI generated and only serve as a starting point for robust discussion in the PR as to whether they will help or not.
Jody took a look at build.geoserver.org failures:
-
docs/en/requirements.txt had been removed:
-
it was being used for mkdocs build via docs/en/pom.xml script
-
This is how release builds and publishes docs across branches
-
something to address in build-release3.sh
-
Please review & discuss!
Underlying failures are not deterministic and difficult to trace…