Executive Minutes of the Geant4 Technical Steering Board Meeting,
4th February 1999
Attendees:
John Allison (Chairperson), Katsuya Amako, John Apostolakis, Frank Behner,
Andrea Dell'Acqua, Peter Gumplinger, David Jacobs, Hisaya Kurashige,
Michel Maire, Maria Grazia Pia, Florence Ranjard, Takashi Sasaki,
Peter Truscott, Marc Verderi, Hans-Peter Wellisch
By telephone:
Makoto Asai, Gabriele Cosmo (afternoon only), Satoshi Tanaka, Hajime Yoshida
Apologies for absence:
Carlo Dallapiccola, Sergey Kotelnikov, Yohei Morita
By invitation:
Gunter Folger, Laszlo Urban
Requests from Working Groups
- Geometry asked Tracking to provide the capability to add a
parameter to be printed at each step.
- It was requested that there should be an example of event biasing.
- Geometry requested visualization
of polygons and the geometry hierarchy (Dtree).
- Documentation requested more manpower.
Software Management
We expect to update PhysicalConstants and SystemOfUnits to the
corresponding CLHEP ones and to migrate to the use of CLHEP RandGeneral, with
the next CLHEP release (foreseen around end of February). Maria
Grazia Pia requested an extension of the functionality of RandGeneral
to provide continuous distributions for use in physics; Gabriele Cosmo will
convey this to the original authors.
The need to define a policy in order to synchronise CLHEP and Geant4 releases
was mentioned, as well as the wish of an easier way of distributing CLHEP,
such as a tar file from the Web.
It was decided to put XVT support on hold and thus not to devote resources to
it.
It was requested that developers reduce the size of some big source files,
for instance those with very long methods, since the compiler does not know
how to optimize them.
Documentation
The first official version of documentation is available on the web; various
working groups mentioned their plans to extend or update it. Tools to improve
the navigation through documentation, examples and code were discussed.
MoU matters
A recommendation will be conveyed to the Collaboration Board, concerning
Gunter Folger taking the role of Software Management Group Coordinator; Gabriele
Cosmo will retain his position in the TSB as SLAC representative and Release
Manager.
Manuel Delfino will replace Juergen May as CERN CB representative.
Peter Gumplinger will ask the CB to include three more TRIUMF collaborators
already involved in Geant4.
Hans-Peter Wellisch observed that some key authors in the hadronic category
appear in other Working Groups; he proposed relaxing the rule of unique working
group affiliation and letting people appear in more than one Working Group, if
appropriate.
It is desirable that the CB should examine applications for new
admissions more frequently than twice a year, at least in simple cases such as,
for instance, the admission of a student of a current collaborator.
It was proposed to rename the Processes and Materials Working Group to Generic
Processes and Materials, better to characterize its actual scope.
A proposal will be circulated by email by John Allison, to rename the Detector
Response Working Group to Detector Response and Persistency, or to create a
separate Persistency Working Group.
It was clarified that:
- User and Category Interface includes the intercoms category;
- the Track category goes with Particles;
- the Software Management Working Group includes the global category and STL
related issues.
On the question of support to users, the Chairman noted in
clarification that the experiments, labs and institutions contributing
to the Collaboration benefit from this automatically by virtue of having
signed the MoU. There is no intent to obstruct the use of Geant4 by
others but the speed with which their questions can be answered may be
limited by the amount of effort available to the Collaboration and
requests for new features will be subject to Collaboration approval. The
CERN representative added that, since Geant4 belongs to the
Collaboration and not to CERN, CERN's signature of the MoU clearly
cannot be taken as conferring membership on all the physics institutes
in the Member States.
Introduction of Granular Libraries
Migration to granular libraries should take place as soon as possible. John
Allison proposed the creation of a working team (John Allison, John
Apostolakis, Frank Behner, Gabriele Cosmo, Hisaya Kurashige, Hans-Peter
Wellisch); the tasks are making sure Geant4 can be built from granular
libraries, disentangling dependencies and evaluating what is needed for users. A
solution was requested for the next TSB meeting.
Migration to STL
It was agreed to proceed through an intermediate migration interface; then
once a working Geant4 version with STL is available, development will
continue with STL only. The time when development is required with STL only
will be defined by the TSB and will be communicated to developers.
Most of the work of building a migration interface has been done by Stefano
Magni as a feasibility study and is available in the current R&D release;
the interface for two classes is missing only. Another person to complete the
work is needed soon. A working team was proposed by John Allison, including
himself, Frank Behner, Gabriele Cosmo and Stefano Magni; the deadline for a
short term solution is the next TSB meeting.
Data Policy
Currently there are three ways to handle data:
- distribution of separate tar files;
- CVS in geant4/data
- static arrays in source code
As there are good reasons for each solution, it was agreed to keep all
the current strategies.
Issue of Patches
It was agreed to continue distributing corrections as patches.
Testing and Release Strategy
Makoto Asai will submit a written proposal concerning quality assurance,
dealing with issues like memory leaks.
It was agreed to have a 6-week deadline for the next release; the deadline
will be the same for all categories. Deadlines for the following release will be
discussed at the next meeting.
It is recommended that before tagging each category runs at least a full
application involving use of that category, on at least one platform.
Date and Content of Next Release
It was agreed to issue release 4.0.1 in May; the target is consolidation,
bug fixes and time performance improvement.
As May (with a prior 6-week deadline for code delivery) is not realistic for
the work planned in the Hadronic Working Group, Maria Grazia Pia and Hans-Peter
Wellisch requested a release at the end of July; it is agreed that this will be
release 4.1.0. It was also requested that a deadline for code delivery
of the hadronic category for this release be of 2 weeks (or less) prior to the
release deadline: this request takes into account the needs related to
a thorough unit and physics testing of the hadronic models included, to ensure
their correctness and robustness.
Categories planning to put new features into the July release are asked to
communicate their intentions by the next TSB meeting.
New Collaborators
A proposal of a Dubna group to join Geant4 was discussed.
Workshop at ESTEC, Amsterdam
It was agreed to hold it the week beginning 20th September.
SPIDER
They are working at requirements for Software Release Tools for experiments
and asked the Geant4 Collaboration to consider whether Geant4 would be
interested in using them. The TSB agreed that Geant4 will watch the
developments.
Miscellaneous decisions
It was decided that the Tracking and Electromagnetic Working Groups
will work at further performance improvements in tracking and some
electromagnetic processes.
Makoto Asai agreed to maintain the examples and to organize the related
working team.
A proposal to have public check out access to the CVS repository was not
accepted; both CVS check in and out are restricted to Collaboration members.
The guidelines of the syntax of command guidance will be issued shortly;
requests to modify commands will be sent to working group coordinators.
Maria Grazia Pia,
18 February 1999