20070910 Tuesday, Core Software Mtg
Attendees: R.Dubois, T.Burnett, L.Rochester, T.Glanzman,
N.Golpayegani, H.Kelly, D.Landriu, I.Pavlin, T.Usher, W.Focke,
E.Charles, D.Chamont
News from Richard:
Recent ETE2 a success although some config issues. First time LAT
powered up from MOC. See new eLog: http://glast-ground.slac.stanford.edu/elog/Index.jsp
55-day run
==========
(Heather)
New GR v11r13 with new GCRcalib (which does not quite compile)
[Update: there are now new tags of GCRcalib and OnboardFilterTds which
should take care of the compilation problems. A new GR will be tagged
later today once the current round of LATEST builds complete.]
(Richard)
* 7920 batch jobs complete (required redoing some number of crashes)
* Vector exception in FluxAlg
* Pointing history files have been produced (FT1)
* Nicola Omodei is processing merit into FT2, adding diffuse response columns,
& creating live-time cubes
Tony Johnson has released a new data portal for SC2 which includes
tabs to the DataCatalog for browsing/selecting data and an interface
to the data skimmer (formerly "pruner" & "peeler"):
http://glast-ground.slac.stanford.edu/DataPortal/?tab=Welcome%20SC2
But...while the interleave code is doing what we asked, it is not
doing what we want. This has to do with the production & interleaving
of the albedo gammas.
Follow-on simulation: Next project is a full sky + background sim of
one complete downlink (3 hours). Will try to get L0 files, pointing
history files (fes), restore pulsars (e.g. Vela), add GRB in FOV,
cosmic ray recon (GCRcalib). Will advertise this run to the community
prior to run to get any last minute requests in.
Short News Items
================
(Heather)
Joanne requested that we make the facilities package build a shared
rather than a static library. This change was tagged v2r14 and does
not seem to be causing problems for GR or ST in the LATEST builds. We
will try to bring this new tag into GR soon.
G4 shared library problem in Windows: Tracy tracked this down to a
misnamed .dll file. Fix will show up in installer soon.
Toby reports getting two versions of Geant from the Installer...?
Navid responds there are two IExternal versions for G4.
New ROOT 5.14g patch to fix bugs. Heather has it running on Win/Lin,
expect it in upcoming GR and the "last" version of EM.
Event Collections
=================
(Eric Charles)
Defining event collections using the new metaRooData package. See
presentation:
https://confluence.slac.stanford.edu/download/attachments/20011/Event_Collections_New.pdf
The idea is to tie together the various data which are currently
spread out across multiple root files (merit, recon, digi, GCRcalib,
etc.) into "event collections" (root files containing pointers to
files and events), and then to provide nice mechanisms for accessing
data indexed in this way.
(Heather)
David Chamont has been working on cleaning up RootIo and has committed
his changes to the HEAD. Heather is testing the code on Windows and
will then integrate the new metaRootData into RootIo to provide Gleam
the ability to read, write, and interpret these new index files.
(Richard)
This will likely be the primary (only) way to access real data at some
point in the future. Monte Carlo may continue to deal with separate
flavors of root files. Still need to sort out how this will integrate
with the DataCatalog. Aim to get this scheme integrated into Gleam in
time for the 1-year run.
(Heather)
A new SLAC GLAST Citrix box up and running, but needs some final setup
before general availability. Heather is admin, needs to understand
how to grant user accounts.
(Richard)
Citrix is a means to remotely access a Windows machine (roughly
equivalent to "ssh" or running an X client under linux) and can be
accessed from a linux machine. This new machine can be used as a
Standard GLAST Windows environment extablished by the Installer
(similar to the noric machines which provide a Standard GLAST Linux
environment).
Python
======
(Heather)
There have been some troubles with the SciTool builds on RM for
Windows due to Python and Swig. Swig 1.3.31 was made available for RM
yesterday after it was found to be missing, causing some of the build
problems. We have also updated to Python 2.5.1 for both windows and
linux. There is an outstanding issue in that it appears a python
add-on for numarray needs to be provided for the windows distribution.
Toby asked if we could move to Numpy since numarray is currently being
phased out. Warren added that he was quite willing to update ASP to
use numpy as well. As Jim was unavailable for comment, we will
contact him via email to see if SciTools can use numpy rather than
numarray. Pending the outcome of that discussion, Navid will then
update the windows python distribution accordingly.
Upcoming Face-to-Face Core Meeting
==================================
Hope to get Toby, Heather and Navid to California to overlap with as
many SLAC folks as possible.
Current schedule: 13-17 August 2007
Skimmer
=======
(David Chamont)
Will release new skimmer after meeting which will solve known problems
and merge pointing history (from merit files) into a separate file.
A technical issue: when using the "fast merge" (merging without
decompression), the max file size parameter is ignored.
(Richard)
Current work of parallelizing the skimming task has encountered an
issue of saturating the file server containing the files to be
skimmed; thus, the benefits of splitting up a single skim into many
dozens of jobs is not as large as expected or hoped. We may find that
a modest parallelization will maximize performance.