Reminder: Please copy/paste the Zoom chat into the notes. If you miss it, this is saved along with the meeting recording, but having it here has benefits.
There was a presentation and discussion about AWS costs, which are significantly over budget. CC decided on an amount to limit the monthly spend, that ACRG or others would figure out how to achieve, and they are writing something up for clarity.
They want to send out a detailed message, haven't done so yet.
mod-marc-migrations; Jeremy was out last week. No news. Need to respect deadline for release. Jason will get in touch with Jeremy.
mod-record-specifications. Evaluation is going on; looks good, don't expect blockers. Ingolf will make a proposal to change the criterion for SNAPSHOT-builds. Julian made a porposal to expand the criterion for consumed APIs. The TCR-45 should be decided on by the TC before WolfCon. Jenn schedules a discussion (next Monday).
Reference Data Upgrades: last week a subgroup was suggested, need to come back to this.
Static Code Analysis: 2024-08-20 Static Code Analysis Meeting notes . The group investigated the current Sonar Cloud settings. Only a few changes to the Default Settings have been made be the development teams. These changes do not apply to the whole project, but have been made on a per-module basis. The group proposes the following set of guidelines:
Default rules for each code scanner per language are the standard.
The team is free to change the rules according to their development needs.
The team is free to set up a new code scanner without explicit TC approval
The changes need to be made explicit to the reviewer and reasonably justified.
The reviewer is a member of the TC who does the code review
Disagreement as to whether one is obliged to use a certain analysis tool for certain projects (e.g. Sonar Cloud for Java Projects). Some say yes, some say no.
Marc: Development teams need more guidance as to what tool and what set of rules to use.
Others say it should be up to the team (and the duty of the team) to choose a tools and setup and to document and justify their choice.
The Pull Request to folio-org/tech-council needs to be worked on before we can decide on it. E.g. discussions need to be closed. The 3% figure for code duplication needs to be dropped if teams are allowed to fine-tune theirselves.
Some say if a new development language comes up, the first team who uses it must propose a code analysis tool and then all other teams must use it (for this programming language). Caveats because projects can have multiple languages and different coding styles. Contradictory to the current proposal that the team is free to set up a new code scanner (even for an existing language).
Jenn: The proposal of the subgroup is not ready to be voted on. There is disagreement among the group members.
The group must meet again. Results will not be brought back to TC before WolfCon.
The Spring Boot support period of 12 months for a single minor release has the same length as FOLIO's support period of ~ 12 months, however, these periods usually don't align.
Regarding Backend "First Party Libraries / Frameworks": A library/framework needs to support only one of the versions allowed in the "Third Party Libraries / Frameworks" list. This should be clarified in the description.
Adding folio-s3-client to first party libraries and the AWS and MinIO clients to third party libraries. folio-s3-client is an abstraction for the AWS and MinIO client.
Java 21
No objections from developers, the migration from 17 to 21 will be easy, simply bump the folio-java-docker image in the Dockerfile and the jenkins-slave-docker image in the Jenkinsfile.
Need to contact DevOps whether they can provide a Java 21 version of folio-java-docker and jenkins-slave-docker before the Sunflower development starts; recent commits (folio-java-docker history, jenkins-slave-docker history) have been made by David Crossley, John Malconian, and Julian Ladisch.
If yes, next step will be to create a ticket for DevOps
Vertx 5
Major release, some changes to how callback, futures, etc.; may affect how we chain things.
Julian has tried and found little difficulty with RMB
Arguments for holding off until Trillium, not want to upgrade both Vertx and Java, but signal this is coming
Note: Trillium expected in Fall '25, assuming no delays; Sunflower needs support for two releases, gets very close to end of Vertx 4 support if there are delays.
Current inclination is towards Sunflower, but not solid decision
For Java 21 & Vertx 5: will wait for Java 21 capacity conversation, make decision next week.
S3/MinIO - S3 API version: still need version, see Action Item below
Previous Notes:
Java 21
Recap:
No objections from developers, the migration from 17 to 21 will be easy. (See above.)
Need to contact DevOps whether they can provide Java 21 environment as above.
If yes, next step will be to create a ticket for DevOps
Note: KeyCloak has already deprecated support for Java 17 and will remove in next version. Support KeyCloak through a Bouncy Castle library. Bouncy Castle support for Java 21 does not mesh well. KeyCloak is aware, could possibly adjust.
Vertx 5
Major release, some changes to how callback, futures, etc.; may affect how we chain things.
Julian has tried and found little difficulty with RMB
Arguments for holding off until Trillium, not want to upgrade both Vertx and Java, but signal this is coming
Note: Trillium expected in Fall '25, assuming no delays; Sunflower needs support for two releases, gets very close to end of Vertx 4 support if there are delays.
Current inclination is towards Sunflower, but not solid decision
For Java 21 & Vertx 5: will wait for Java 21 capacity conversation, make decision next week.
S3/MinIO - S3 API version: still need version, see Action Item below
OpenSearch 2 & Elasticsearch 8:
Some differences have crept in.
FOLIO currently uses only the basic features thought the OpenSearch client that are available in both.
Marc Johnson has run into some more basic issues with Elasticsearch client in another project, more strict.