Choose timezone
Your profile timezone:
ROOT Team Meeting
ROOT Team Meeting
Zoom Meeting ID
63815034196
Host
Danilo Piparo
Alternative host
Philippe Canal
Useful links
Join via phone
Zoom URL
ROOT Team Meeting 2026.07.27
Liaisons report
-
ATLAS: Presentation on histv7 by Jonas at AMG on Friday
-
ALICE: We had to revert deprecation/removal of TGrid* family of classes following official request by the experiment.
[Vincenzo]: Would they be open to having the classes deprecated in code with annotations?
[Giacomo]: Not discussed yet, if we manage discussing soon we are still in time for changes before 6.42
- LHCb: there is some new effort on the integration of RNTuple in DaVinci, Silia is following up on this.
Round table
Jonas Rembser
- Delivered presentation about HEP analysis on GPU.
[Danilo] Great presentation, could be used as a baseline for more
[Jonas Rembser] Some people were asking me to use some of the visuals in the SFT induction
[Jonas Hahnfeld] I was surprised by some messages given in the talk, and the presentation was about ROOT. Maybe next time slides could be shared earlier?
[Danilo] Any take-home messages for the project?
[Jonas Rembser] After Bryce's talk I was thinking I would like to use the tile programming approach in RooFit, but I am yet to try it out concretely.
Jonas Hahnfeld
- Developing collection unrolling for RDF+histv7
[Vincenzo] Good feature of RDF, otherwise it would be inconvenient for users having to write a custom Define.
[Jonas Hahfneld] That wouldn't even be enough, you would need a custom action.
[Vincenzo] Yes, indeed.
Jake
- Could produce a root-dataframe wheel working on top of root-core wheel.
Two main issues to address:
* I needed to remove the modulemap from the wheel, which gets rebuilt by clang
* I need to manually patch the modules themselves
[Jonas Hahnfeld] Yes those were my two issues I hinted at last week
[Stephan] Where is the modulemap written by clang?
[Jake] We have to write it to site-packages
[Stephan] You may not always have the right to do so
[Vincenzo] Yes, we are aware. Going more in general than Jake's work, I currently believe that a promising way forward for other issues related to the wheels is to require to have write permission to site-packages. For the moment, I believe this issue should not block Jake's work
Bertrand
- ROOT 6.36 can't be installed with more recent versions of VS
[Giacomo] I also found that it can't be installed with latest clang on Linux
[Jonas Hahnfeld] It's because of a new libc++. I think this is fine, 6.36 is an older release and we usually state that we don't support new platforms on older releases.
Anna
- I will present a poster next Monday, will add a link later.
AOB
[Stephan] About TGrid, we discussed at the ROOT workshop that when we deprecate classes we allow a long deprecation period for the community to adapt. Then we did again what we did in the past and removed a class that was just deprecated in the stable release in master. I believe we should decide on a deprecation period and keep it long.
[Giacomo] This case was particularly different from other times, we didn't actually deprecate the classes but moved them to the Deprecated namespace. We should also decide on a strategy. Ideally they should have been marked as deprecated but not moved to a different namespace and then left there for at least another version.
[Stephan] I believe deprecating for only one LTS cycle is still quite strict, it usually means just one year and we know our community doesn't always have the time and dedicated effort to address deprecations.
There are minutes attached to this event.
Show them.