Description

bp-alex

opened on Nov 7, 2025

Issue body actions

The Change Log shows one critical fix (#252) plus a significant number of performance regression fixes ready for release in 0.9.2.

I think we should release 0.9.2.

In respect of our open PRs and tickets:

• I commented on Remove MDB_UNSIGNEDKEY, add a builder for Dbi, add MDB_INTEGERKEY comparators #276 and suggested we defer it to 0.10.0 and provided reasons there
• Regarding Iterator performance #270, I think similar comments would apply, but as it is marked as a draft I am deferring reviewing it for now. I also believe we should merge Remove MDB_UNSIGNEDKEY, add a builder for Dbi, add MDB_INTEGERKEY comparators #276 before Iterator performance #270, as Remove MDB_UNSIGNEDKEY, add a builder for Dbi, add MDB_INTEGERKEY comparators #276 touches on a wider surface area and Iterator performance #270 would be easier to follow if Remove MDB_UNSIGNEDKEY, add a builder for Dbi, add MDB_INTEGERKEY comparators #276 was already in place
• My only concern is CursorIterable used with BACKWARD_AT_LEAST and BACKWARD_CLOSED is not returning expected ranges of results when using DUPSORT #267, but I see this will get fixed as part of the broader refactoring in Iterator performance #270. As such I
  don't mind making an exception to my usual "zero bug policy" in this release, as CursorIterable used with BACKWARD_AT_LEAST and BACKWARD_CLOSED is not returning expected ranges of results when using DUPSORT #267 is not being ignored - the fix is just being deferred for semantic versioning reasons and there is already code that fixes it

As such, are we happy to release 0.9.2?

Comments, @stroomdev66, @at055612?

Activity

1. stroomdev66 commented on Nov 7, 2025

   

   stroomdev66

   on Nov 7, 2025

   Collaborator

   More actions

   I agree with all of the above comments. I'm happy for 0.9.2 to be released.
2. at055612 commented on Nov 7, 2025

   

   at055612

   on Nov 7, 2025

   Collaborator

   More actions

   Releasing 0.9.2 is fine by me. I agree that #276 is too big to go in a patch release, especially since it deprecates various methods and removes the MDB_UNSIGNED_KEY flag.

   I think we will combine the #276 PR with the fix for #267 into a single PR as both depend on each other to fully test their respective changes.
3. bp-alex commented on Nov 8, 2025

   

   bp-alex

   on Nov 8, 2025

   MemberAuthor

   More actions

   Thanks @stroomdev66 and @at055612.

   I've released version 0.9.2: https://repo1.maven.org/maven2/org/lmdbjava/lmdbjava/0.9.2

   When running the release plugin, I set the next version to 1.0.0-SNAPSHOT. This keeps each version number segment to a single digit and signals the API changes proposed in the open PRs by moving to a 1.0.0 major version.

   Regarding the pending changes: could you coordinate on breaking your commits into smaller, reviewable units? The current 12,000+ line scope makes thorough review and potential future bisection challenging. I understand the work requires some large commits, but separating out discrete pieces where feasible would help considerably.
4. bp-alex

   closed this as completedon Nov 8, 2025

Issue actions