- 3 openlab students to work on SyCL and GPU support in RooFIt, and ??
- Enrico wrote the report on Shush; a bit of backlog in the forum
- Bertrand takes over this week
- I/O: most important remark is that we were discussing the size differences in RNTuple w.r.t. TTree for the ATLAS PHYS format (RNTuple is half the size)
- TMVA/RooFit: ??
- LIM: same state; nothing interesting to mention
[no topics planned to discuss this week]
Jonas R.: contacting a GSoC student who did very well in the warm-up exercise (to work JSON I/O in RooFit). Also working with Garima, and some stuff to be finished for CHEP (about GPU support, etc.)
Vassil: a bit of work on cppyy on top of clang-repl; it's early work so almost nothing works. Also some paperwork for a GSoC student that maybe is moving to Princeton. Some work on the paper on Automatic Differentiation for error estimation.
Philippe: on vacations last week. This week I hope to address some bug reports from last week.
Baidyanath: further work on PR (some work related to Numba?); cppyy test failing because of missing features. Also some work on template support for libinterop and `enum`s support in cppyy.
Sebastien: progress [me taking the minutes unable to catch some OpenGL-related technical stuff]. Meeting with Matevz. I will work in render pass management (e.g. layers).
Florine: making the plots about the RNTuple/TTree size comparision for ATLAS PHYS, for the LHCC presentation. Remark: considerable size difference in the uncompressed size (which was discussed with Jim Pivarski on the last I/O meeting). This week will contrast the explanation, maybe moving to also take a look to performance.
Jakob: EP R&D day, SW package Plan of Work seemed well-received; LHCC slides. Also met with Oksanda/Alexander regarding Analysis Grand Challenge benchmarks. This week: improving tests for RNTuple column encoding and also working in bulk I/O.
Dante: talking with Lorenzo and Vincenzo on how to proceed with the batch generator benchmarks. maybe a PPP can be scheduled at some point on this topic.
I was going to have a meeting today with Jakob and Javier for the RNTuple-related CHEP project, but we deferred util tomorrow.
Bertrand: mostly working on the new CI: setted up a new Windows machine with the latest Visual Studio version. Trying to contact Ole to solve some issues.
We had a problem with C++ modules after updating to the last VS version on Windows. I will check with Vassil, but this is not sustainable if it breaks after each update.
Also, I will be on shift this week.
Olivier: taking care of some posts in the forum, and some issue (Nan, inf) in graphs -- errors were not visible from a Jupyter notebook. I also fixed Doxygen file for v6.26.
Finalizing work on log-scale `TAxis` drawing.
Garima: last week working on paper (and also submission for artifact review). Also some work on a RooFit PR.
This week: proceedings for ACAT, and discussing next steps with Jonas.
Vincenzo: 2 PRs open: (1) one in `v6-28-00-patches` to integrate the things that were missing for matadata handling; (2) the other on avoiding re-JITting on distributed tasks. A few discussions with the rest of the team.
Discussion with Andrea Schiaba about inconsistencies on I/O throughput (local vs. remote via xrootd).
Javier: meeting with Marcin from ATLAS a couple of times: one already mentioned by Florine to show the results on RNTuple/TTree size comparision; the other to discuss news on our side, e.g. regarding late schema extension. Also had a meeting with HPE to discuss current issues and future use of their cluster. Applying to get access to Oak Ridge's Summit too.
2 PRs opened last week: foundations for late schema extension in RNTuple and improvements over the collection proxy support.
This week: mostly ACAT proceedings, but a small PR related to RNTuple also coming.
Lorenzo: discussing with Vincenzo about the batch generator, and also some review that I was discussing also with Vincenzo. Also working with Sanjiban ob GNN.
This week: making choice for the openlab Summer Student.
Axel: gone last week. Soliciting feedback on the Open Source programme. Selecting/interviewing the next generation of fellows.
This week: ensuring that all bugfixes in `master` have been backported and fixing some TTreeReader issue with Double32_t.