- GlastRelease status: (Toby, Richard)
v10r4 has been
made and has been used already to produce some all-gammas, however we need a
new version of OnboardFilter
in order to run with multiple filters concurrently, in particular CNO
and gamma. That code is ready, but, because of other enhancements
(support for DFC filter, including periodic triggers) which
are not needed immediately but will be with time, it
depends on a new version of
the obf external library. After some effort,
Navid has managed to build it on Linux. He doesn't yet have a Windows
build; Tracy may be able to shed some light on the process.
[post-meeting update: Tracy provided a Windows build of obf.]
(Richard) Navid has recently switched RM over to a new, dedicated
account, glastrm. However, until recently, this account was not entitled
to the same priority as the glast account in batch, so RM builds have
been proceeding at a snail's pace. Now things are back on track.
There was some discussion of whether everything is in place to
handle CNO properly, e.g. where/how to prescale, proper simulation
of deadtime... Leon regrets that the heavy-ion people are expecting
to use this coming run without mods to tracker code to make clusters
wider. (Richard) We'll be doing this again in about June.
Beamtest (an interlude topic): (Leon)
Heads up that we may want to upgrade; it's currently based on GR v9r22,
far enough back to make certain things awkward.
- GR and the 55-day run: (Richard)
See this checklist. (Tom) Preparations for Pipeline are coming along.
There is currently a glitch accessing ACD calibrations [Joanne has provided
instructions for dealing with it; longer term ACD calibrations need to
be rationalized, but that will have to wait until Eric Charles returns (soon).]
(Toby)Why are we using real calibrations for this run? Is there a
significant difference? (Leon) There is some loss of efficiency due to
dead strips; also ToT distributions change. (Berrie) The dead strips
will affect energy recon, which estimates energy in the tracker based
on hit strip count. (Leon) The number of dead strips is well under 1%
except for the thick layers of one tower. (Berrie) What about noisy
strips? (Leon) There is no way to simulate behavior of noisy strips.
We assume they are masked off and treat them as dead. (Julie) It is
important to use the same calibrations for background runs as for
all-gamma. (Richard) Toby will probably need to redo his runs.
(Julie) suggests we do a small run with calibrations first
and examine before proceeding. (Richard) In effect this happens anyway:
Tom makes the early part of a run available for examination.
(Toby) would like to see the effect of the somewhat more realistic
spacecraft geometry. (Richard) No one has looked at it at all; too
risky to incorporate it in this run. (Toby) It's not as risky as it
might be since it doesn't involve any detector volumes.
[..but meanwhile Bill discovered a bug in the ACD doca calculations
so we'll have to start over with the 55 day-run checklist when Eric returns.]
- CMT Action Committee report (Heather)
There has been considerable activity concerning a couple of
relatively small-scale issues:
- We'd like to turn off the automatic definition by CMT of many
environment variables. Navid is in the process of implementing
an alternative which will eliminate the need for complicated
environment setup before running our executables.
- Jim requested that Tracy's fix for building shareables on Windows
be incorporated. Tracy did some research and discovered that a newer
version of CMT, v1r18p20061003, also has the fix. Although the official
Glast version is at v1r16pxxx, some (e.g. Joanne) have been using
a slightly earlier (than the version with bug fix) version, v1r18p20060606, for nearly a year. Heather
and Joanne built GR v10r4 on Windows and Linux, resp., with the new CMT.
It looks ok so far. We also need to verify that it fixes the problem it's
supposed to fix. (Toby) Can we make the new CMT version available from
the Installer? (Richard) Let's wait until we make the decision to use
it – or not. (Heather) It's available from the
CMT download area.
[Post-meeting update: Navid plans to try out the new
CMT for LATEST builds only.]
Meanwhile discussion continues on the larger issue of what course
to take with regard to CMT: do we patch it, ditch it, or something
inbetween? We will be considering
various alternatives
- Gaudi-slim (Heather) GR is practically
up and running on RHEL4, but there is one stubborn problem remaining:
a crash on exit from some test programs which seemed to have something
to do with one or both of
gaudi and CLHEP.
She made several changes
to gaudi to eliminate references to
various other packages which, by default, gaudi
pulls in, such
as Seal, BOOST,
xerces, ROOT.
(Our uses of xerces and
ROOT are not mediated by
gaudi; the fact that
they are by default pulled in
by gaudi has potential for causing
confusion if we're not using the
same version.) Normally she would prefer to leave external libraries
untouched, but in this case there were substantial benefits:
- Mysterious crashes went away!
- The Windows gaudi
build we distribute is 160 Mbyte; the neutered
version is a mere 40.
J. Bogart, Last Modified:
01-Jun-2010 15:48:28 -0700