Reference Guide Updates
Cassandra has been working for over an year on a restructuring and revamp of the already awesome Solr Reference Guide. The intention is to merge these changes in and release with Solr 9.0. There are a lot of changes which have been broken down in to multiple commits, but those all belong together and so would be merged in one go. Cassandra is going to call out for help when she’s ready for the rest of us to pitch in and contribute.
Solr 9.0 Release
1. Cleanups and deprecations - we should continue to work on those. Alan and David worked on some of it in the past, but we should continue to clean this up so we can have a good 9.0 release. Currently we have code that was supposed to be removed in prior versions, and we should take the time to clean that up as much as we can.
2. Lucene release blocks Solr 9.0 due to the dependency. We should start discussing and work towards a Lucene 9.0 release so we can plan Solr’s release accordingly.
3. There were some discussions around Solr and Lucene development process, something that has also been discussed on the mailing list.
4. Plan to move the Solr code up one level, and clean up things a bit on that front were discussed along with cutting of a 9.0 branch in the near future, but it was decided to wait until closer to the Solr release, so we minimize the number of branches to back port code to, and also the difference between main and 8x is kept as little as possible.
5. Discussion around “if we should recommend everyone to back port” as much as possible, so that it’s easier for others to cherry-pick/backport from main to 8x branches. It was decided to leave it at the discretion of the individual committer but reasonable number of people thought it made sense to back port more rather than less, as 8x line is very much alive and 9.0 still has some time to go.
Committer Update: Ilan
Ilan shared the scalability improvements he’s been working on and the intention to optimize Solr to only load and use the resources that are really required. Removal of the single Overseer node was along those lines.
A lot of us have different tools we use to benchmark changes, though none of them are complete or ‘recommended’ solution. Attempts to reduce tech debt doesn’t really allow the tracking of performance at this point, leaving committers with less confidence around such changes. The idea to converge on such tooling if possible was discussed though no concrete pointers or directions were noted.
Mark was present at the meeting and shared his thoughts on the ref branch. There aren’t any folks actively working towards trying out his branch at this point, and with little or no support in the future, that may not be recommended as well. Mark mentioned he intends to get inspiration from the work he’s done and continue to pull in stuff from there, but doesn’t have active plans to work on that at this point. Other folks noted how the branch has also worked as an ad-hoc reference while fixing performance issues on the main branch. People are free to take inspiration from the branch and bring some parts of that into main as they like.
Community appreciation and being thankful
The call ended with everyone appreciating the community, and each other about all the work they’ve done for the project (code and community). A lot of us spoke about and appreciated things like but not limited to Jan’s effort around creation of the TLP, Houston’s effort on the operator, Cassandra’s work on the reference guide, and every one else’s effort on so many other things that have made Solr the project it is.