meeting 26th May
Fabrizio Furano, Sebastian Binet [REMOTE], Peter
Mato, Predrag Buncic, Gabriele
Cosmo, Benedikt Hegner, John Harvey, Fons Rademakers.
Line Management Issues (John)
NeuroBayes license (Pere)
Report from the LIM meeting (Benedikt)
News from projects
version 1 (8/6/2011 @ 15:10)
Management Issues (John)
- The fellow for the AIDA project will be starting 1st July
under the supervision of
- Start a
negotiation with NeuroBayes for a new clause
granting access to the source code
under certain conditions (Pere)
answer received after informing them that only LHCb will be signing the
AF meetings frequency (all)
- Discussion and positions from the experiments:
- CMS feels that the AF meeting is far too frequent. Dropping it
once/month or less frequently would be more appropriate.
- ATLAS is happy with the current fortnightly meetings. A meeting
every month would also be OK, but less frequent than this would be
detrimental for the
- LHCb position is exactly the same as ATLAS.
- ALICE is also in favor of one meeting per month.
- Predrag mentioned that the current frequency was adequate for
the R&D projects. They were supposed to report at the meeting and
this was driving the progress in many occasions.
- It was agreed then that the AF
meeting will take place once a month on a fix time and weekday slot
(Thursdays at 14:30) and a constant phase within the month. The
proposal is to reserve the second Thursday of every month in
- The e-mail channel will be used for urgent matters or decisions
that cannot wait for the meeting to take place.
- The frequency of LIM meetings will not change.
Plans for the completion of the ROOT schema evolution (Fons)
- The work for the completion of the ROOT schema evolution has been
broken down into a number of tasks, and their required
effort has been estimated. Fons will send the list of tasks to the
AF mailing list and will be also added into the minutes ==> Action Fons.
- Fons claimed that the CMS use case is more a 'model evolution'
than a 'schema evolution'. For this case what is needed are some
handles in which to plug the CMS code that will perform the model
conversion. The first task of the list to be implemented will be
[number 6?] for the benefit of CMS and Fermilab will put the manpower
for it. After this, Philippe will be working for the rest of
- Pere asked whether there is a test case provided by CMS to
validate the desired functionality. Fons said that this is not needed
in this case since the ROOT and CMS teams will be working together on
- The tasks will added to the JIRA planning to monitor progress. ==> Action Pere.
Report from the LIM meeting
See the minutes
of the last LIM meeting for
- The LCGCVS to CVS migration is considered done.
- CMT. Benedikt complained that the release of the new CMT
version was not done correctly. He will follow this
with David R. Also LHCb didn't have any chance to give feedback before
release was done. ATLAS had a preview in the nightlies. It was felt
transparency in the CMT development is needed.
- SL6. A test cluster has been provided by IT. Questions such as
how we do the migration, which
compiler version has been started to be asked.
- The system compiler is gcc 4.4 and is probably adequate,
but this probably contradicts a previous discussion that we should go
to gcc 4.5.
- LHCb reports that a build with gcc 4.3 does
not work in SLC6. What would be nice for them is to latch on a version
of the gcc compiler that works on both OS versions. In this
case gcc 4.4 would work fine for both SL5 and
- The argument for gcc 4.5 was the better vectorization support.
... LHCb will like to
start testing soon.
- We should take a decision on the compiler version at next AF
and we need to prepare for this ==> action Benedikt.
- Benedikt will be collecting feedback on the SNOW (new CERN
service support portal).
- Python. Started to see the need for version 2.7, which would
be useful for later migration to 3.x. LHCb is not planing to make a
News from Projects
- CernVM (Predrag)
- ATLAS distributed computing WS last week.
- It was demonstrated Panda jobs on CernVM and having the
output to EOS.
- IT+ATLAS agreed on a plan for an evaluation including
performance compared with more traditional methods, total
cost of ownership, efficiency, I/O options, memory
over commits. Timescale of a couple of months.
- A new version of CernVM is under testing. It will be announced
at the next
- LHCb with DIRAC is standing in the queue ...
- CernVM-FS. Last week ATLAS started running production.
- ROOT (Fons)
- Next week a release candidate for 5.30 will be produced.
Fons explained the new process that has been put in place: there
will be two
release candidates before producing the final production release.
- The trunk will be copied into the release branch.
- It was agreed at the LIM to start using
the release candidates in the nightlies.
- Most of the planned 'stories' are going to be there (e.g. new
user guide, first
version of CMake to build the windows version, etc.)
- Review of the included 'stories' at the AF next meeting.
- Preparing the monthly release. This will be the last one before
release in June.
- Successful first running of the validation using CernVM-FS,
currently limited to the CERN site.
- In the coming weeks they will be asking for a new LCG patch
configuration that would include a patch for ROOT with
a fix of a floating point exception.
- Initial SLC6 discussions are taken place in the collaboration.
What will be needed to help them is some symlinks
in the external area packages to allow them to move forward. ==> action
- CERN and the C++ standards
committee (John Harvey)
- CERN is now a member of the Swiss Association of
Standardization (SVN), which is the Swiss
part of ISO IEC JTC 1 SC 22 (contains the C++ Standards Committee).
Because it's CERN (and not a single person), we can send anyone we
want, and we have a vote that can be cast by whomever CERN sends. This
membership give us also access
to documentation free of change.
- Next meeting in 2 weeks on June 9th.