Present: Joanne Bogart, Toby Burnett, Sasha Chekhtman, Xin Chen, Johann Cohen-Tanugi, Richard Dubois, Heather Kelly, Traudl Hansl-Kozanecka, Michael Kuss, Sean Robinson, Leon Rochester, Alex Schlessinger, Karl Young
See also Toby's Core Issues/Status document for additional comments on several of the items below.
Releases: v4r1 is the very last "old Gleam" (in the sense that it uses Gaudi v9 rather than v11, builds an executable, etc.). It compiles and runs on Linux as well as Windows.
Tools status: (Toby) See this page devoted to migration issues, primarily but not exclusively those relating to Gaudi v11r1.
Shipping zip files of the new stuff among the various developers proved to be extremely awkward. Copies of all relevant packages now exist in the CVS repository under the top directory GlastRelease. This works much better and should be considered for future collaborative development of this kind.
There are currently two things holding up development under vc7:
Some of the Tkr classes in reconRootData are not handled well by rootcint, resulting in output that does not compile under vc7. One workaround is to revise the Tkr classes somewhat, which Tracy believes he can do with no special difficulty.
Most external libraries do not need to be recompiled for vc7, but G4 does. Norm Graf has provided access to his instructions for necessary patches and compilation with vc7.
(Michael, et. al.) Transition to gcc
3.2 is stuck, pending resolution of a stubborn runtime
failure, a series of error messages followed by a
segmentation fault. Certain of the errors have been eliminated
by inserting pragma statements in strategic
places, but not the ones having to do with iterators. One theory
was that the problem was restricted to ROOT applications, but
that seems not to be the case. Test programs for packages not
dependent on ROOT, such as GuiTest.exe and test_FluxSvc.exe also
fail with a seg fault. Test programs, at least some of them,
with no dependence on either Gaudi or ROOT run properly. We
could try sample programs supplied with Gaudi to see if the
problem already exists there. Some things Michael has already
tried which didn't help:
Upgrading to latest gcc (3.2.1)
Making changes to Gaudi and HTL to get rid of compiler warnings
We need to try building everything with gcc 3.2 elsewhere, both to verify that the problem is not a result of some procedural error or environmental problem at Pisa and to get more people involved in this effort. So far Alex has built everything from scratch on his Linux box and has seen the same problems as Pisa folk. He will copy his build to public SLAC Linux so that more can play.
(Traudl) See extensive notes on her work on a parallel approach to making Gaudi an external package.
Calibration: (Joanne) The calibration infrastructure design has been revised somewhat: data to be retrieved from a persistent store will undergo two stages of conversion, one corresonding to the metadata and the second for the bulk data (XML or ROOT file). For details, current status and a to-do list see the Gaudi and Calibration Infrastructure page. Highest priority is to assemble all the pieces needed for a working example. Once this is done, it will be time to nail down details of data design for the different calibration data types with the subsystems. In addition to demonstrating the viability of the infrastructure design, the example will serve as a model for how clients may access data.
EM Data Unpacking: (Heather) She has made her way through various M. Huffer documents on the EBF (an AUM: Acronym of Unknown Meaning) data and has some relevant code, written by Selim, who adapted it from something JJ wrote. Somebody from Flight Software might have the time to provide a nice class interface for us, but we can't count it.
Action Items
previous
|
minutes index
|
next
|
J. Bogart Last Modified: 01-Jun-2010 15:48:25 -0700