A new document regarding how ROOT responds to analysts' needs will probably be written.
We should be releasing soon. Large backport of RooFit commits is merged. Probably the release will happen next week.
Currently only one blocker https://github.com/root-project/root/pull/13134 . It seems to be a problem with TClass generation of dictionaries for transient std::unique_ptr data members. Needs debugging and it would be nice if it happened soon.
Progress on the benchmarks with distributed RDataFrame on the HPC cluster at CERN.
Finishing off the RDF progress bar effort. Continuing with the systematic variations tutorials.
Working on RooFit. Research on NVIDIA GPU to optimize time execution. Troubles with building ROOT.
Working on integrating functionalities of code coverage for ROOT.
Working on the debug builds of Windows. Apparently on Windows if you want to build debug, every part of the software stack must be built with debug symbols. MSVC becomes also extremely picky with debug builds, which uncovers a lot of previously unseen bugs.
Working on the upgrade to LLVM16. One problem with MacOS builds is building on /tmp, which is limited in size and triggered unclear issues. CI is being moved to build on the home directory.
Working on the poster of the summer student project.
Finished PR for on RNTupleInspector.
Working on support for std::set, found issue in TVirtualCollectionProxy. Philippe pointed out correct way to use the class, Javier updated documentation.
Starting CHEP proceedings
Discussing with Jolly about the SYCL implementation, doing research on the subject.
Improving ONNX operators support for SOFIE.
Worked on fixing segfaults due to data races in core/meta (affecting RBatchGenerator and distributed RDataFrame)
Supervision with Silia, Dante, Joseph
Found inconsistencies between TClassEdit and TDataMember.
PR review
Supervision with Victoria, Florine
Improve documentation for TVirtualCollectionProxy
Support for custom IO rules may introduce some overheads, working on fixing them
Trying to work with the linear algebra parts of ROOT. Looking at BLAS/GSL to find interfaces for operations on matrixes.
Project: parsing deep learning models from external libraries into SOFIE. The parser needs to be in Python.
Working for simple networks, need to test with more complex operators.
Later on, moving to decision trees with sklearn and xgboost.
Compare JSON outputs of pyhf and ROOT. Results are similar but not exactly the same. Need to check how pyhf runs the fit in the background.
Support for enums merged in RNTuple.
RField API streamlining done.
RNTuple chains in RDataFrame's RDataSource.
Bulk I/O patches for RDF brought into RNTuple library.
Preparation of the PPP meeting, brainstorming follow-up.
Setting up new workstation.
Registered for student presentations, in two weeks for now.
Bureacracy, catching up.
Aware of the new CI causing problems.
Sonatype lift will be ripped out from the CI since it's being retired.