ROOT team meeting 03.03.2025

News

1

Danilo will present the status of ROOT as a project to the LHCC tomorrow, 04.03.2025.

• Giacomo: I believe the work on the RNTupleMerger was far more difficult than foreseen and would warrant an extra item in the PoW
• Stephan: when are we allowed to increment the completion of an item in the PoW?
  • We try to be conservative and just increment by 0.5 an item completion, but we can decide on increases of 0.25 case by case

2

Silia Taider joins the ROOT team as an ORIGIN graduate, working on Python interfaces and interoperability.

3

On the 24/25th of March we will have a ROOT internal retreat in IdeaSquare. Team members are invited to propose topics for the work to be done during these two days.

Shift

• Issue with IPython and metakernel came up during the weekend. Identified and fixed by https://github.com/root-project/root/pull/17862.

Meeting summaries

I/O meeting

Discussed about the RNTuple validation suite. Currently getting contributions from Jonas and Florine.

PPP

• 20.02.2025: Patrick Connor presenting Darwin ntuple production framework in CMS. Advanced usage of TTree and friends, outstanding issues tracked on github.
• 27.02.2025: Tomas Dado presenting FastFrames ATLAS ntuple production framework for analysis. We currently cover most of the performance functionalities required, but there are still some nice to haves.
• 06.03.2025: Mu2e analysis framework status and future plans.

Misc

• Danilo: can/should we reduce the number of MacOS versions we support on our CI? The motivation could be to have more bandwidth in the Mac nodes to unclog PR builds.

• Dev: A draft PR I opened is just removing patches, everything seems to work and performance seems unaffected.

  • Vassil: Possible test to check one of the removed patches. Load a file once, unload file, change content of file and reload it again.
  • Vassil: in some cases when we read from PCH somethings are not serialised. If you have a macro and you want to check macro definition, you have to touch disk. I think in the past we put in place optimisations to avoid having to touch disk in such cases.

• Danilo: do we want to run the RNTuple validation suite regularly?

  • Jonas Hahnfeld: Don't know yet.

• Giacomo: I'd like to merge the update to the CI that enables the dev=ON option. Similarly I would propose to suggest to use dev=ON for their builds.

  • Jonas Hahnfeld: The usual reason not to do that is that -Werror is prone to false positive warnings from compilers. It would steal more of my time than benefit the project if I had to enable this configuration.
  • Giacomo: Ok, I have not had the same experience yet, but we can discuss at another time.

• Vassil: Regarding the unavailability of support for CUDA + modules, one other option is to ask NVIDIA to do it since lately they have increased their effort in LLVM/Clang.

• Jonas Hahnfeld: It would be good to understand the timeline for the branching.

  • Danilo: It was highlighted in a previous status report doc, I will also update today's status report. Link to previous report https://cernbox.cern.ch/external/eos/user/d/dpiparo/Documents/SFT/ROOT/WeeklyStatusReports/2025/250224_weekly_Status_Report.docx

• Stephan: how will it work for the integration of roottest in ROOT?

  • Danilo: I will take care of this.