Present: Torre, Liz, Robert Fay, Benedikt, Michel, Dario, Pere, David
Pere - went to the “Future of S&C in NP” workshop in JLab. With Amber did some advertising of HSF. People were positive in the discussions. Encouraged people to join in Paris.
Registered participant count is 20 so far.
We discussed the agenda with the following outcomes. Michel will update indico accordingly.
We will invite volunteers to bring guinea pig projects to an HSF project incubator/template hackathon on Monday morning, coordinated by Benedikt with the assistance of Pere. If there is a critical mass of takers, we will hold it; if not, we won’t. We need to produce a sufficiently clear and detailed description of what it will involve to attract the sort of people who should be interested.
Opening session - Mon afternoon
No new additions to the guest speaker list. We should now lay out a draft agenda for this session as best we know now.
Possible talks added in the discussion:
NB Trying out Jupyter/ROOT requires CERN login, should inform people in advance how to get a basic CERN account.
Machine learning - Tue 9am
Organizers: David Rousseau, Jean-Roch Vlimant
A Fermilab person is also in the loop to help organize.
First ideas on session content, thanks to David for the notes:
RAMP hackathon, a second half-day, may be a possibility, sounds interesting (need to decide a specific topic, quite some work to prepare). Would need more information on what the hackathon would be and then poll whether there is critical mass of interest.
Given the possibility of a second half-day it makes sense to have the ML session on Tuesday morning, then if there is a hackathon it will follow Tuesday afternoon.
Note: need to update ML web page on HSF website. Clarify relations with IML. [David et al should just give us revised content, we’ll be happy to update.]
Packaging - Tue 2pm
Organizers: Liz Sexton-Kennedy, Benedikt Hegner (or designees)
We agree we’ll have this session, oriented towards introducing the community to Spack [which has achieved the rather remarkable feat of uniting the diverse participants in the HSF packaging forum – each of whom previously had their own favorite solution ;-) – in a belief that Spack is the best basis for a HEP common solution] and addressing work needed to adapt it for HEP use. Benedikt just listed some issues on their discussion list. See also our hep-spack github repo.
Geant4 technical forum - Tue 4pm
Organizer: Makoto Asai.
A suggestion from the discussion: follow up on what was discussed at SLAC that it would be useful to have a configurable software stack that Geant4 can use with its releases to avoid (at least for HEP-targeted releases) baking into the release particular sw versions.
[Post-meeting from Makoto:
Our technical forum at Orsay will be 90 minutes, 45 minutes of introductory presentations by Geant4 developers on recent new features and near future deliverables, followed by 45 minutes of users’ presentations on their status/issues/requirements. This technical forum meeting is of course open to anybody including HSF members who have nothing to do with Geant4, to see as a model case how a software project would interact with its user community. We expect user participation would not be limited to HEP but also from nuclear physics, space and medical applications, material science, and if we are lucky from security. ]
Software concurrency and performance - Wed 9am
Organizers: Vincenzo Innocente & other draftees
A proposal for this session was discussed and agreed.
Closing session - Wed 2pm
Benedikt - packaging TN is finished, modulo any final input from Liz.
Liz signs off.
Liz - Spack work continuing, studying how it can evolve to meet our needs. Participation from Fermilab, Benedikt in his spare time etc.
Pere - meetings & discussion take place where?
Liz - context is Spack developers meetings, these are open, announced on Spack dev Google group.
Benedikt - LHCb person is interested in building the stack up to Gaudi with Spack.
Dario - end result of CERN sprint is there are now ~50 developers working on the platform. Becoming rich with training material for universities. It is now time to inject our requirements for HSF and LAL workshop. eg. navigating easily through related courses.
Riccardo will be present in Paris, prepared to offer hackathon with volunteers willing to learn the platform in real time.
Liz - there’s an FCE person dedicated to HPC training materials. Any way to leverage WikiToLearn?
Dario - collaborative effort is the main point, but also we will be adding private pages with controlled content. So stable training material can be added by others under their control. Can also be used as a portal to other materials, e.g. there are portals to materials controlled by university professors.
Michel - please ping Riccardo about updating our website WikiToLearn description.
Robert - OpenEDX(see
http://open.edx.org/) should be complementary to WikiToLearn, production ready, Liverpool looking at hosting it, should examine whether it looks useful for HSF. Should have a prototype link by next week. Looking at it for our intro course for postgres, to provide the course on demand.
Dario - agree with looking at this additional platform, can be complementary. Ditto for others such as Software Carpentry.
Pere - a lot of progress made recently on Jupyter/ROOT notebook pilot. Will draft guinea pigs after Easter, expected to be production ready in June.
Dario - no new entries. There are presently about 6-7.
Michel - perhaps send a reminder.
Voting should proceed prior to the workshop. Dario will check what we said about timeline.
Nothing to report.
WLCG report to write. Should have HSF content. Benedikt will pass Torre some guidance on what to write, it is just a paragraph or so.