Minutes (written by A. Ribon) of the LCG Generator Services
planning meeting on 27 November 2009.
----------------
People attending in room 32-1-A24:
- Gabriele Cosmo (GCO)
- Gloria Corti (GC)
- Lev Dudko (LD)
- Judith Katzy (JK)
- Mikhail Kirsanov (MK)
- Pere Mato (PM)
- Witold Pokorski (WP)
- Alberto Ribon (AR)
- Alexander Ryabov (ARY)
- Lars Sonnenschein (LS)
People attending remotely via EVO or telephone:
- Jon Butterworth (JB)
- Claire Gwenlan (CG)
- David Grellscheid (DG)
- Peter Richardson (PR)
- Torbjorn Sjostrand (TS)
=====================================================================
AR presented the status of the ongoing activities in the LCG Generator
Services subproject since the last planning meeting (29 May 2009),
and the plan of work for the next 6 months.
Comments to the slides
======================
1) Comments to slide #6 ("Migration to autotools 1/2")
a) JK said that also ATLAS is very interested to have Mac OS X
builds for most of the GENSER generators, because a large
number of users have Mac laptops and would like to be able
to run some MC generators.
b) AR replied that so far Mac OS X builds are available only
for few generators requested by LHCb (Alpen, LHAPDF, Photos,
Hijing, Pythia6, Pythia8, Herwig, and Jimmy).
For other generators, like Herwig++, we encountered some
problems when trying to build it for the platform:
osx105_ia32_gcc401
c) DG commented that he builds regularly Herwig++ on Mac OS X,
without any problem, and for different versions of the
system (e.g. 10.5 and 10.6).
d) MK said that we have not investigated very much on the
problem found when trying to build Herwig++ on Mac OS X,
but it seemed a configuration problem of the Fortran
compiler installed in the system...
He agreed that we need to look more in detail, and eventually
get some help from DG to fix the problem.
e) GCO suggested that we should add a new milestone on GENSER
regarding the migration for all generators to Mac OS X,
including the 64-bits architecture.
f) AR and WP agreed on the migration of Mac OS X, but commented
that this could take a considerable amount of effort, which
should instead be dedicated to testing.
We all agreed that we should at least try to do the porting,
but not investing too much effort if problems will appear.
------------------------------------------------
2) Comments to slide #7 ("Migration to autotools 2/2")
a) JK requested that it should be clearly documented how to use
the "bootstrap" script, because it can be very useful for
users.
b) AR agreed, and said that at the moment the documentation is
quite short and a bit dry, therefore it needs to be improved.
------------------------------------------------
3) Comments to slide #8 ("Progress report: Validation")
a) MK commented that MC@NLO has now a test, and work is in
progress to have one also for POMWIG.
b) DG asked whether the GENSER tests used for each generator
are shipped with the tar-balls, in order to let the authors
to check whether they are sensible.
Indeed, recently one of the Sherpa author (Frank Siegert)
found that the test used by GENSER to test Sherpa was not
appropriate.
c) MK replied that the tests are available from CVS, but they
are not shipped with the tar-balls, because they are meant
to be used only by GENSER developers, not by users.
d) JK commented that she was able to find some of the GENSER
tests from the GENSER web page, without major efforts.
e) WP suggested that, as part of the review of tests used by
GENSER (which is one of the major activities proposed for
the next 6 months), we should inform each MC author of the
tests that GENSER is currently using to test her/his
generator.
------------------------------------------------
4) Comments to slide #9 ("Progress report: HepMC")
a) GC suggested to make an addition to the "LCG nightly builds"
in order to let also the experiments to make some tests with
the newest version of HepMC.
This requires the availability, for at least some generators,
of the libraries built with the latest version of HepMC
(instead of those available on AFS, that are built against
the old version HepMC 2.03.09).
b) PM required, as one of the crucial tests of the new HepMC version,
the possibility to read back HepMC files generated with an older
version of HepMC (in particular with HepMC 2.03.09).
[ After the meeting, Lynn Garren informed us that
"These tests are routinely included in the HepMC test suite.
We all recognize that they are very important.
But, of course, only the ASCII format is tested since
this is the only standardized output." ]
c) GC supported PM's request, and added that it should not only
be tested with ASCII files, but also with binary ROOT files.
d) Related to c), PM added a major request for HepMC: it should
make available an extra library that allows binary I/O with ROOT.
This would not introduce any extra (ROOT) dependency in the
main HepMC library.
PM reported that the Linear Collider community asked several
times how to do binary ROOT I/O with HepMC, while each LHC
experiment does it in a different way, integrated with their
framework.
[ After the meeting, Lynn Garren commented this request
as follows:
"This raises a number of issues. First and foremost,
HepMC is supposed to be a standalone package with no
dependencies. So a ROOT I/O library would need to be
completely separate and NOT built by default.
I don't even know how to decide which version of ROOT to use.
That said, people have been asking for this capability for
some time. They are also asking for non-ROOT binary I/O.
We'll see what people say at the meeting - tentatively
scheduled for Feb. 3." ]
e) DG said that d) should be discussed at the HepMC meeting.
f) AR accepted the requested a), b), and c). For d), it will be
put as an important item of discussion for the HepMC meeting.
------------------------------------------------
5) Comments to slide #10 ("Progress report: MCDB")
a) LD commented that several of the CMS requests have been already
addressed, and others are in progress.
LD informed us about a new version of the HepML library that has
been released recently.
------------------------------------------------
6) Comments to slide #13 ("Proposed milestones")
a) As reported above (see 1-e and 4-d), two additions have been
suggested:
- Migration of all generators to Mac OS X, including
the 64-bits architecture.
However, we should not invest a considerable effort on
this migration in the case of major problems.
- Discuss the possibility to add an extra library in HepMC
to allow binary I/O with ROOT.
These have been included in the revised version of the
presentation (see slide #13).