Notetaker: Jonas R.

News

  * Ioana is joining the ROOT team, remote for now. She has been working on awkward 2.0 and moved to uproot some weeks ago. Now she is also trying to get into ROOT.

  * discussion with openlab to get access to xeon accelerator for decompression

  * Axel is with Microsoft to get acces to azure for dist RDF demo, also for RTuple storage side

  * discussion with NVidia on compression and decompression algorithms on the GPU, also with Ana

Shift report

  * handover from Axel to Lorenzo

Meeting summaries

  * IO: discussion about different C++ types in RNTuple, like enums, unique_ptr etc. These are all connected to experiment AODs. Talking together with IT to test RNtuple together on a larger scale, but still in early stages. This week, xrootd workshop where Jakob wants to figure out if we use xrootd in an optimal way.
    * Somthing with time that I was too slow to type here :(
    * Late schema extensions: renamed to "late model extension"

  * LIM: discussion on C++20 issues. ATLAS have a 6.28 blocker (there is a GitHub issue, something with PyROOT). ALICE is happy. But they really want to have C++20 support in the next weeks. They want to migrate to C++20 at the same time as migrating to GCC 13.

Regarding C++ 20

* Vassil is working on the modules part, supposed to be done by this week
  * conditional header files in the modulemap files is the big challenge in keeping C++17 support e.g. with GCC 5
  * Axel: why do we need a single modulemap? Vassil: otherwise it becomes a maintainance problem.
  * We should also revisit the GCC version we support at some point

* Part 2 is we also need to work on our code to support C++20
  * in C++20, you can create a `string_view` from an iterator of strings. Therefore, `{str1, str2}` becomes ambiguous, because it can both resulve to `std::vector<std::string>` or - as of now - `string_view`.

ROOT swag

* Poll results: still have some time to vote.

* T-Shirts: people prefer the simple black and white shirts

* Hoodie with logo instead of slogan is preferred.

* Stickers are all great.

* Axel remembers we will have old mugs with signatures, 10 or 13 CHF for one

* Betrand notes that the T-Shirts should not be too cheap, otherwise they break after washing two times

Going over GitHub issues



Going over the JIRA issues


Roundtable

* Jakob
  * type support for rntuple `std::bitset` `std::unique_ptr`
  * slides for the xrootd workshop
  * gone Wednesday to Friday

* Vassil
  * mostly GSoC
  * finalized ACAT paper
  * boost in cppyy by using pointer aritmetics and not strings
  * using C++ modules in cppyy

* Lorenzo
  * Working with Sanjiban on sophie (GNN)
  * Minuit benchmarks for CHEP
  * More TMVA work

* Garima
  * Pushed initial implementaiton for RooFit code generation and minimizer interface

* Ioanna
  * worked on uproot issues

* Olivier
  * made the new TScatter class
  * invenstigating when ROOT on macOS was still building with X11, it didn't work at least since 6.14

* Serguei
  * several new features in JSROOT, like XY-projection
  * trying to close all known issues with web canvas to make is the default soon
  * in vacation from 2nd of April
  * Somewhere end of April wants to make the web canvas the default

* Philippe
  * waiting for PR reviews

* Sebastien
  * Rander Q (queue?)
  * Rendering pipeline customization
  * Has two more months

* Bertrand
  * a bit of CI (debugging with Ole), in particular on Windows
  * having fun with spaceship operators

* Florine
  * set up basic experiments to measure ATLAS Physlite throughput
  * got Feedback from Javier and Jakob on her PRs

* Jolly
  * getting fixed and variable binnings on GPU
  * going to fo

* Jonas R.
  * worked on RooFit AD with Garima, first RooFit code generation in ROOT master
  * worked on slides for the LHCb talk tomorrow in person, finalizing these today
  * contacted the summer and GSoC students -> openlab student now confirmed
  * job shadowing high school student from May 23

* Giovanna
  * went back to coding on the S3 backend
  * Working on TTree to RNtuple which still takes a bit too long
  * investigating long term issues with memory registration, related to reusing the memory for the pages

* Jonas H.
  * Submitted abstract about ROOT on RISC 5
  * Debugging reprodible failures when having specific build directly paths

* Axel
  * Follow-up work to the openlab workshop
  * Spent quite some time on the new CI
  * C++20
  * this week: more openlab and Microsoft meetings, more CI and C++20