Executive Minutes of the Geant4 Technical Steering
Board Meeting,
25th March 1999
Attendance
In person: John Allison (chairman), John Apostolakis,
Frank Behner, Andrea Dell'Acqua, Gunter Folger, Vladimir Grichine (on behalf
of Sergey Kotelnikov) Michel Maire, Youhei Morita, Maria Grazia Pia, Florence
Ranjard, Marc Verderi, Hans-Peter Wellisch
By video: Makoto Asai, Katsuya Amako, Hisaya Kurashige,
Takashi Sasaki; Gabriele Cosmo; Peter Gumplinger
By telephone: Hajime Yoshida
Apologies for absence: Carlo Dallapiccola, David Jacobs,
Satoshi Tanaka, Peter Truscott (due to technical difficulties with
teleconference link)
Minutes
The minutes of last TSB meeting were approved.
Granular Libraries
They are now implemented and tested on Linux and Sun; a few technical details
will be discussed among Frank Behner, Gabriele Cosmo and Gunter Folger.
It is expected to have granular libraries in the May release.
It was decided to have the granular option, as well as the current
"global" one in place for the May release, and to consider an additional option
of intermediate granularity, requested by Gabriele Cosmo, for The July release.
The Geant4 Collaboration will recommend the use of granular libraries,
but it will be up to the users to decide wich option to adopt.
STL migration
Frank Behner reported that with a few days' work he completed the STL
interface, resulting from Stefano Magni's feasibility study. Now a Geant4
version that compiles and links with STL is available; this did not require any
change of code in the whole of Geant4. Frank estimated a week's work was
necessary to complete all the testing and debugging.
CB and TSB reports
John Allison presented a brief report from last CB meeting; further details
are available in the CB Executive Minutes.
It was agreed that the TSB chairman will submit TSB reports to the
TSB prior to presentation to the Collaboration Board.
Update on Italian participation to Geant4
As a result of last CB meeting the Italian Collaborators appear as "Italian
Geant4 Developers" in the new version of the MoU. A new IGD Collaborator
- Alessandro Brunengo, with affiliation to the Hadronic Physics working Group -
has been admitted at last CB meeting.
Giovanni Pieri was accepted as a new member of IGD, affiliated to the
Geometry Working Group.
Report on Electromagnetic Physics
Michel Maire reported on the recent activity of the Electromagnetic Physics
group; a new LowEnergy directory includes the developments related to the
ESA low energy project; a test directory includes various comparisons between
Geant4, Geant3 and EGS.
Report on Geometry
John Apostolakis showed a detailed report on the recent activity on Geometry.
Alternative constructors of replicas are being considered for a future
release; the change would be trasparent to users. A problem in tracking
in a field (observed in the Navigator by Gabriele Cosmo) will be fixed
in time for next release.
g3tog4
The issue of support of g3tog4 was discussed.
Report on Hadronic Physics
Hans-Peter Wellisch presented a detailed report of the activity in progress;
people have been allocated for all the items planned for the first half
of the year. Contacts with the Durham group have been fruitful, as he had
got access to several experimental data sets through them. Work is in progress
to set up a test suite at the microscopic level for hadronic models. Contacts
are also in progress with a theorist from Genova.
Geant4 performance
Geant4 geometry is much faster than Geant3.21 geometry; it is expected that,
even where more complex physics models have been implemented in Geant4
to produce more accurate results, Geant4 physics will not be slower than
Geant3.21 one. It was agreed that more detailed evaluations of physics
performance and accuracy should be made available by next TSB meeting, to
complement those already presented in the RD44 final report.
Test Procedures
Improvements of the test procedures are planned for the July release.
Serguei Sadilov is working at an improved tag database.
Maria Grazia Pia encouraged people to post results and comparisons with
physics data on the Geant4 web Gallery.
John Apostolakis invited John Allison to present a detailed System Testing
schedule; he requested a 48 hour System Testing turn around and suggested to do
statistical tests submitting long jobs on weekends. John Allison requested to
let the System Testing team decide how many tags to submit together to System
Testing, in consultation with taggers.
Policy of Patches release
It was agreed to issue correction patches at most every two weeks; their
content and frequency will be agreed with Working Group Coordinators and
taggers. A patch will go through the System Testing before being released.
Patches will only contain bug fixes, not any new features. It was also
agreed to issue granular patches regularly and to group some of them later.
Libraries
It was agreed to distribute libraries in optimized mode for the supported
platforms with each release.
Gunter Folger asked for feedback about shared libraries, as LHC++ is
moving to them. It was decided to discuss it offline and to have a report
on this topic at next TSB meeting.
UserAction Classes
Following a discussion initiated by Michel Maire about the uniformity of
UserAction classes interfaces, a proposal by Makoto Asai for uniform const
pointers in the methods of all user action classes was accepted. This will
break users' code, so it was decided to have the new interface along with
the current one in the May release, and to advertise that the current interface
will become obsolete starting from The July release.
Random number seeds
John Apostolakis reminded the TSB that the next CLHEP release, including
functionalities requested by Michel Maire, was scheduled around the end of
March.
Gabriele Cosmo could not directly participate in this discussion, due to
technical problems with the teleconference link, but had provided a note
confirming that the problems concerning random numbers reported by Michel
Maire have been fixed in the new CLHEP Random module since last December and
that the new CLHEP release was scheduled for early April.
Makoto Asai illustrated how the Random Engine is handled by the Event
and Run Managers and volunteered to provide the ability to turn on the facility
to store the status of the Random Engine in the RunManager.
Gunter Folger informed that CLHEP release procedure is quite complicated,
as it must go through CERN registration rules.
CLHEP
Gabriele Cosmo reported about the CLHEP meeting held at Fermilab on March 1-3,
where representatives from FNAL and CERN were present as well, Gabriele
representing SLAC/BaBar and GEANT4. Among the topics discussed in the meeting,
also issues related to collaboration among labs/groups contributing to CLHEP
were discussed. In this context, the GEANT4 project, as also part of the
LHC++ strategy, has been identified as a "special client" or "Collaborating
Project", which essentially means that a certain priority to G4 needs is, and
will be, given in CLHEP. It has been stressed the necessity for Geant4 to have
a well defined procedure and schedule for CLHEP releases in order to easily
syncronize the migration to new versions of CLHEP with the Geant4 releases.
Code contributions from Geant4 to CLHEP have been discussed (G4Allocator
classes) and a collaboration of Geant4 experts in the new STDHEP project is
foreseen.
Release Plans
Next release will be on 27th May; the deadline for code delivery is on
16th April.
The following release will be at the end of July; the deadline for code
delivery will be defined based on the experience gained with the May release.
Maria Grazia Pia confirmed that, resulting from discussions within
the Kinetic Model group at an advanced state of the OO design, a realistic
deadline for the delivery of their code would be around mid July.
Quality Assurance
Makoto Asai suggested finding more people to run memory leak checkers and
invited category coordinators (or people appointed by them) to do code
inspections in their categories regularly.
Hans-Peter Wellisch and John Apostolakis commented that the output
generated by Code Wizard was too noisy: a few relevant messages were
interspersed through many useless ones.
StdHep
Makoto Asai is in charge for following developments with StdHep; he had
nothing new to report.
Web pages
Gunter Folger reported that new web pages are in place and invited the
TSB to give feedback before posting them to the public. It was decided
to set up a team (John Allison, John Apostolakis, Gabriele Cosmo, Gunter
Folger, Maria Grazia Pia) to finalize the new web pages. Working Group
web pages will be optional to each Working Group. Maria Grazia Pia suggested
to have links to member institutes/experiments/labs Geant4 web pages, as
used to be the case from the RD44 Geant4 homepage.
User Support Services
John Allison reported that an automated bug report system had been put
in place by Serguei Sadilov. The TSB expressed the appreciation of Serguei's
work. As a further test within the Collaboration, prior to public release,
it was decided to do a turn around within the System Testing Team for a
global tag and to report the outcome to the TSB by email; if stable, the
tool will then be released to the public. The System TestingTeam and the
Software Management Coordinator will verify the stability of the tool before
its public release.
Documentation
Concern was expressed about the future of LIGHT II; therefore it was decided
to continue with Stefano Magni's documentation tool, which the current
Software Reference Manual is based on.
MoU updates
Hans-Peter Wellisch requested modification of the Responsibility Matrix to take
into account members of the Hadronic Physics Group contributing also to
other Working Groups.
Policy of document availability
It was decided to continue issuing Internal TSB Minutes (for TSB distribution
only) and to issue public Executive Minutes.
The communications concerning tags go into the tag database.
Mailing lists
It was decided to have an announcements list, that is a one-way list, to
notify interested people of some major pieces of news, such as the announcement
of the availability of a new release.
There was a consensus that technical discussions of common interest should
use the widest appropriate list.
Workshop
It was decided that next Geant4 Workshop will continue as in RD44 style,
that is open to Collaboration members and some invited users.
Next TSB meeting
Proposed dates are 31/5 or 10/6; consultation will follow by email.
Maria Grazia Pia,
30 April 1999