# GEFF Java Meeting
September 25, 2025
Attendees: Caroline Malin-Mayor, Ko Sugawara, Tobias Pietzsch, Jean-Yves Tinevez
## Spec v1 is out!! :octopus:
- Most relevant change is different spec for strings: relying on zarr-extensions var length utf8-encoded string spec
- https://github.com/zarr-developers/zarr-extensions/tree/main/data-types/string
- This is implemented in python already and is similar to what we were planning
- Plan is to implement it in the n5 library and not in geff
## Roadmap to GEFF Java v1
- Getting geff-java to the level of the python library is a substantial effort, will lag behind
### Step 1: Patch the current implementation so it works for TrackMate and Mastodon to support Geff spec v1
- Variable length property implementation - there but might need to be slightly adapted to meet the spec changes - Ko with Caroline reviewing/helping as needed
- Trackmate - missing polygons - Jean Yves
- Mastodon - missing ellipsoids - Ko
- Coordination with N5 library on variable length string implementation
- [x] Caroline TODO: make an issue on n5 repo and discuss async, set up meeting with zarr/n5 folks if needed
### Step 2: Get funding for Tobias to generalize the code
- Equivalent of core_io in python library
- Java geff representation needs some additional features to ensure dtype fidelity when reading and writing
- Maybe also missing features that aren't needed for TrackMate/Mastodon but needed to fully implement spec
- Jean Yves is working on a grant/collaboration that would include GEFF-Java work, but it will take a bit
- [ ] TODO: Caroline talk to Preibisch about OSSI finds for Geff Java
### Other Logistics
- Jean-Yves suggests we could try Nature Methods
- the original cell tracking paper format was a nature methods paper and no one even used it
- Cross-talk between Java and Python tools is a big selling point
- [ ] TODO: Caroline set up next meeting beginning of Nov (Ko) and/or end of October (Jean-Yves and Tobi)
- a month is not a lot of time so it will just be a check-in, not a progress report