Core Minutes 10/12/2004
Present: Joanne Bogart, Toby Burnett, James
Chiang, Johann Cohen-Tanugi, Pol D'Avezac, Seth Digel, Richard Dubois, Warren
Focke, Marco Frailis, Riccardo Giannitrapani, Berrie Giebels, Navid
Golpayegani, Traudl Hansl-Kozanecka, Heather Kelly, Michael Kuss, Julie McEnery, Chuck
Patterson, James Peachey, Alex Schlessinger, Tom Stephens
-
Announcements (Richard)
- There is a LAT managers face-to-face meeting on Wednesday (solely
face-to-face; laptops are banned!). Expect the latest news on schedules.
- David Chamont will be visiting SLAC for a week, probably in November, to
work on CalRecon revamp.
- ScienceTools:
(James) Tag is imminent. There is a last problem or two to be ironed out
to get all the unit tests to run. Should the tag wait for the upgrade in
documentation James is about to do? Consensus was to go ahead and tag as soon
as the code works. James will add to the Doxygen documentation over the next
few days. As soon as there is a HEAD version which is a candidate for tagging,
Toby will check it out and attempt to build on Windows.
Julie wondered
whether there should be some instructions for non-SLAC Linux users on how to
install Python 2.3. This is a more recent than the version which comes
natively with supported Linux platforms. Turns out Python is currently only
used by the sane package for running System
Tests, not something typical users would do. Furthermore, it just invokes the
Python executable to run scripts. There is no dependence on Python libraries
and no requirement for a particular version of Python, so there is no need for
a Python external library at all. This could change in the future if we use
BOOST.
- Infrastructure requests:
- Can the RM make Doxygen documentation for HEAD builds as well as for tags?
At some point might also want to consider adding more to HEAD builds, such as
System Tests. (Navid) Should not be difficult to make Doxygen for HEAD builds.
- What happened to GlastRelease v4r5p1? It doesn't show up on, for
example the
GlastRelease builds page. (Alex) This is fallout from the AFS and batch
system failures. The v4r5p1 build is there, but the RM web pages need to
be correctly regenerated, which he will see to.
- Can Installer handle tags as well as LATEST? (Navid) Yes. In the short
term, he will do the necessary work manually to make the upcoming ScienceTools
tag (v5r0p1?) available from the Installer.
- How about installing the Installer binaries, etc. here at SLAC so that
people here can run executables without CMT? (Navid) It's already done.
See the bin subdirectory of an RM build at SLAC, such as
/nfs/farm/g/glast/u05/builds/rh9_gcc32/GlastRelease/GlastRelease-HEAD1.366/bin.
A typical file in this directory is a script which first defines environment
variables, etc. as needed, then runs an executable.
- GlastRelease: (Toby) He has revamped
IExternal requirements files recently. ExternalLibs
now just defines the EXTPACK_DIR macro and nothing else. Elsewhere in
IExternal child packages set has been expunged in favor of macro
wherever possible. He also disabled the InstallArea for the time being; it
can't be really put to use until we move Gaudi versions, which is not likely
to happen soon.
A little more is known about the memory leak reported in
JIRA issue TRG-3.
Leon has traced it to OnboardFilter. There are some uninitialized variables
which could be partly responsible, but it's hard to believe they account for 9 Kbyte/event.
∫ Flight(w, ..) dw
- EM status:
(Heather)
v3r0404p2, made last week, is still the latest tag, but she's about to
make a new one including new tracker and calibration tags, among others. Would
be nice to get the ROOT file headers in, but there are still build problems on
Windows. In light of Toby's IExternal changes a little fix is needed to
ldfReader.
- ToT calibration: (Richard) Path for
calibration data to get from Online to us is still uncertain. Johann's first
strategy was to collect the input data for his algorithm from memory, before
it gets written by Online and write a FITS file as output. He then used
astroroot to make an equivalent ROOT file. However this means that the
GlastRelease package CalibSvc, which reads in the
file, has to link to astroroot, undesirable since it's not well-supported.
Next proposal is to use zipped xml as the format. A sample 16-tower file he
made is 20 Mbytes zipped. [After the meeting Joanne did a little testing. It
takes ~2 minutes just to parse the unzipped file with Xerces. This is
unpleasant, albeit not a show-stopper; memory usage is another potentially
worrisome issue. Stay tuned for strategy #3 next week!]
-
Pipeline:
-
(Richard for Dan) Last week's
comments ("Goal for this week is to finish off proper
archiving of successful runs and notification when there are failures") still
hold. These things were trickier than expected.
- (Richard for Matt) Working on pipeline front-end editing.
Coming Soon
- TkrRecon revamp presentation next week.
J. Bogart Last Modified:
01-Jun-2010 15:47:54 -0700