Key4hep Discussion

→ Europe/Zurich
Vidyo

Vidyo
Live Notes
Notes Archive
Hide

# Key4hep Live Notes

This is a document for taking notes during Key4hep meetings.

Date: April 21, 2020

Agenda: https://indico.cern.ch/event/909291/

Connected: Weidong, Jiaheng, Clement, Andre, Erica, Placido, Valentin, Tao, Frank, Gerri, Xingtao, Pere, Paolo, Marko P

Apologies:

## Communities Round Table


## Build system / Target compiler / dependencies

## Podio

* retagged v00-10-01
    * Version in CMakeLists.txt

## EDM4hep

* Tagged EDM4hep

## K4FWCore

* Tagged v00-01

## MarlinWrapper

* nightly ilcsoft build on top of a custom view of LCG_97
* /cvmfs/clicdp.cern.ch/iLCSoft/lcg/97/nightly/
  * x86_64-centos7-clang10-opt/
  * x86_64-centos7-gcc8-opt/
  * x86_64-centos7-gcc9-opt/

* Create a layered release for Key4hep, to be discussed via PM

## Distributed Computing, Workload Management, Data Management

* Important when workloads have to be run
* Add VOs to existing instances ilcdirac, ihepdirac, ...
    * Develop common DIRAC plugins

## Questions on "software deployment workflows" for a Pre-GDB meeting

* https://indico.cern.ch/event/813800/

1) The Key4HEP platform: where does "your" software stack start and
   which platform do you expect to be present (e.g. CC7/8 with heposlibs,
   LCG externals, only singularity to start you own container)

    * Containers used for CI jobs to provide CC7 for example
    * Software based on, e.g., LCG97, which depends on heposlibs
        * (LIM discussion on HepOSLibs slimming, April 28, 2PM)
        * some fluidity in the boundaries
    * (Can also install dependencies via spack)

2) Your use of containers: do you expect the grid or the batch farm to
   run (singularity) containers for the detector studies?
   
   * No fixed decision either way, will do whatever is necessary or convenient
    
3) Is software deployment for production different from software
   deployment to developers?
   
   * Yes, developers will do what is most convenient.
   * Different platforms for dev and prob
   * Obtaining dependencies and deployment should be the same for dev and prod

## Gaudi

* Discussion with Marco Clemencic
    * Agreed that we could base ourselves on cmake development branch
        * https://gitlab.cern.ch/gaudi/Gaudi/-/merge_requests/986
        * https://gitlab.cern.ch/clemenci/Gaudi/-/tree/cmake-modernisation
    * Will be part of Gaudi v34r0
    * To be deployed in ~~dev3~~ dev4_Gaudi, keeping other versions stable
    * dev4 with Gaudi v33r1p1 with patch for interface libraries
     
* CEPC
    * Based on gaudi v32, will migrate to v34
    * Based on LCG releases

* Once dev4, dev4_Gaudi available see how to proceed
    * Maybe create LCG_97_key4hep with our preferred version of Gaudi

## TODO

* CVMFS: /cvmfs/sw.hsf.org
    * /cvmfs/sw-nightlies.hsf.org
    * Need automatic deployment
        * Spack configuration added
        * Using Gitlab CI
        * gitlab.cern.ch/key4hep
* Discuss spack with SPI/HSF Packaging working group
    * Spack recipes in k4hep-spack

* Define Convenors, key4hep-sw-admin

## Done

* Spack: Instructions are on the website


## First release

* Content:
    * EDM4hep
    * K4FWCore: dataservice, +++
    * Marlin Wrapper
    * DD4hep plugin for EDM4hep output
    * Example
        * TODO: Simulate a few muons and reconstruct with MarlinWraper
* How to tag releases
    * using release script from iLCSoft for individual packages

## AOB

* E-group key4hep-sw will remain open with admin approval


## Next meeting
May 5, 2020

## Call for Logos

 

There are minutes attached to this event. Show them.