HSF Startup Team meeting notes, Nov 26 2014

HEP Software Foundation Startup Team Meeting notes Nov 26 2014

https://indico.cern.ch/event/355255/

Present: Torre, Andrea, Dario, Daniel, Pere, Michel, Frank, Pete, Graeme, Brett

Meeting notes:

==== News on contacts made, discussions had, ...

Torre - the community mailing list now has 94 members. A long way to go.

Graeme - talked to Andrew McNab, gridsite author, about vac/vcycle, a project for dynamically provisioning VMs. He is interested in engaging with HSF, will come to SLAC, a good candidate for a presentation from a small interested project.

Dario - difficult to get feedback within INFN. Making direct phone calls, trying to build interest.

Torre - haven't heard back from astro contacts yet

Had a discussion about openness to projects that are small and/or not among our explicit focus areas; in complete agreement that we are open to proactive interested people/projects, including at the workshop. An advantage of having a lightning session, can hear from many.

==== Contact meeting plans

Contacts thus far have been small/one-on-one discussions, and presentations to groups/experiments. We agreed previously that once we have our document draft out, we should start planning more organized (still informal) discussions: announced meeting, indico page, agenda, meeting notes, etc. Such meetings are complementary to the small/one-on-one discussions.

*** We put together a list of organized meetings we will try to have before the holidays, with responsibles to make the contacts and try to organize a pre-holiday meeting if there is sufficient interest:

Generators/theory: Pere
ATLAS: Torre
CMS: Daniel, Pete
Linear collider: Frank
LHCb: completed (Pere)
ALICE: Pere
Intensity frontier: Daniel, Brett
Astro: Torre, Dario
Belle II: Pere, Torre

Points from the discussion:
- GENSER mailing list can be a way to reach the many generator projects, at least for LHC
- we should also keep in mind talking to our close cousins nuclear physics (beyond ALICE)
    Michel - can discuss through French channels, eg. with Panda
    Torre - will review Jefferson Lab S&C early next year, will discuss with them
- Dario - talking to astro particle community in Italy. Interest in participation. CTA experiment needs overlap with HEP. Interested to hear about HSF, contribute ideas. Talked to smaller community at FNAL, test beam infrastructure. Interested in DAQ, as is a community in Italy. Will be represented at SLAC.

==== SLAC workshop

There has been much good discussion fleshing out the SLAC workshop agenda. Discussion continued in the meeting, particularly on the software project visitors we might invite, notable Apache.

*** We agreed to invite Apache rep(s) to participate, aiming to have the relevant session at the end of the first day, specifics to come.

We agreed that the Apache Software Foundation (ASF) looks potentially like a good model for the HSF, it will be very interesting to hear from them and form our own conclusions. Any conclusions we're able to come to can be part of the wrap-up discussions at the end of the workshop.

*** Pere and Torre will put together a draft agenda based on the discussion.

Torre - Recent word from Richard is that the workshop announcement approval had been held up internally at SLAC, it has now gone to DOE for approval, expected soon (no earlier than next week).

*** Our goal was to provide enough information (workshop agenda, official web page, expectations for ancillary meetings) by the end of November to allow people to plan their trips.

==== Questionnaire

Daniel circulated earlier a proposed list of questions for the pre-SLAC questionnaire, which framed the discussion:

1 In general terms, what should the HSF be/do and not be/do?
2 Any feedback related to the work performed so far by the startup team?
3 Could you comment on the white papers summary document?
4 How could the HSF be helpful (not helpful) to your project?
5 What should the next steps for the HSF be after this workshop?

Pete - we should also ask people what they can/will bring to HSF. What collaborations they are willing to form.
Applies both to software projects and experiments. e.g. in CMS have been discussing which things would we be willing to collaborate with other experiments on, via the HSF. More important than what HSF should do or bring is what people/projects/experiments bring to it; the former arises from the latter.

The discussion recalled that an action we promised after putting out the draft was that we would put out a list of proposed activities/services for the HSF, distilled from and implied by the draft. This list will inform the questionnaire, in addition to the document draft. Andrea circulated a draft list.

*** We agreed we'll put together this list as another input/background to the questionnaire and discussions.

Brett - FCE working group in the US is working on a report, a vision for software, much of it applies to HSF. Have something like a market, new experiments can say what they need, learn how to meet their needs. A mechanism could be an incubator, providing a forum and expertise for formulating problems that are shared by others and may have existing solutions, get guidance on solutions, and/or on finding funding to develop solutions, etc. An interactive function, seeking out people who can provide help.

Pere - yes, like a consultancy team, group of wise people who can give advice to new experiments. It complements the function of making available info to allow people to make their own decisions (knowledge base on the website).

==== Brett's wiki experimentation

Torre recently invited Brett to prototype another interesting approach to building the knowledge base on our website. While Drupal together with rather a large number of plugins provides the key functionality (eg. cross referencing software used among experiments), it may not be the best way. Mediawiki (the software behind wikipedia) with its semantic extensions is another approach Brett is exploring. The semantic extensions add the notion of the wiki behaving like a relational DB. A wiki page can be used much like a relational database object. The HSF content types implemented in Drupal like experiment, project, etc. can map onto semantic wiki objects. Can do queries on them, form links between them, etc. If an experiment uses ROOT, on the ROOT page you can have a query that lists all the experiments using it. 
Can also connect to external sources like LDAP, software repositories, etc.

Brett hopes to soon expose the prototype at hepsoftwarefoundation.org. It may be well suited to a complementary role to Drupal, with Drupal providing the slick front page etc., and semantic wiki the engine for the knowledge base. We shall see, we will all be able to compare and comment when it is out.

==== AOB

We will have our regular meeting on Monday.