Core Minutes 10/25/2011ScienceTools: (Jim) No milestones have been reached in the last week but there is some ongoing work. Johann and a student have been working on understanding the biases seen in the unbinned analysis for diffuse sources, and I have been consulting with them on that. In addition, I have made further tests on the new energy dispersion implementation for the binned analysis and have been posting results on the confluence page describing that work. Once I'm happy with the tests, I'll send a note out to everyone to try out the alpha version of the code.
FSSC: (Tom S.) No news.
Pass7: to reprocess or not to reprocess: It's still up in the air. (Leon) Art may have sorted out issues concerning backgrounds; at least now he has all the facts. (Toby) He is concentrating on something — improving PSF above 5 GeV — which is not terribly relevant to analyses. PSF there is already good enough. Improving PSF below 5 GeV would be of more interest.
GR/Pass8 (Tracy) Some time was spent at the meeting last week to see where we are w.r.t. the schedule advertised at the time of the last International Finance meeting. Another one is coming up soon (Nov. 4-5) at which Bill will give a presentation.
One scheduling issue to be decided is when to work on truncations. The changes for truncations would have many repercussions elsewhere which would delay the release of Pass 8 significantly if done before.
In other news, Philippe Bruel is doing good work on energy recon above 1 TeV. And there is an unresolved issue concerning overlays.
L1 Proc (Heather) There was a bug in reconRootData, now fixed. It can be put into production as soon as we have systests results.
More systests (Heather) The goal is to create SCons and GR builds from the same set of tags a set of tags which differ from a production v17 release only in files that relate to the build system, then compare this release with the production release. However in the first attempt the systests results were different; apparently some significant code differences crept in. Heather is making another, more careful attempt. (Leon) suspects problems running systests are due to Computer Center outage(s). Some things may need to be removed, reset, etc. [Later Leon reported that the systests did complete.]
GR and redhat 5 (Heather) The Computer Center would like everyone off of Red Hat 4 by Dec. 9th. We might make it. She needs to finish up making a Red Hat 5 Gaudi build. And we'll need to validate.
ST and redhat 6 (Heather) All necessary externals are built. Kim ran into a snag with python, specifically with the matplotlib unit test. In order to get it to work he had to drag in other things, which dragged in still other things.. The consensus is that we don't need to add to the distribution purely to get that unit test to work.
More rhel5 migration (Joanne) M.E. has plunged into the job of converting all the svac code to build with SCons. She is either done or very close to it.
obf on Windows (Joanne) is working on building obf on Windows using SCons. Some parts build all right, but she is stymied by the dynamically-loaded libraries. The linker is complaining about an unresolved external reference. Insofar as it's possible to tell what VS does when it builds, she's attempted to duplicate compile- and link-time options used by VS when creating CMT builds in the SCons builds but so far, no luck.
SCons Windows RM (Tom S.) The Windows builds continue to lose track of where the config data is, and when this happens, we get no information because there is no log. He's looking into ways to get diagnostic information out.
SCons Windows RM (Tom S.) Recently a build failed because it just ran out of time. There is a timeout specified to Qt in the code (apart from timeouts imposed by lsf queues), normally sufficient, but this time it was exceeded. He will increase it and will see about upgrading RM code to put out an error when this happens.
AOB
Tom S. will be off tomorrow (and the rest of the week) at a LIGO workshop. They're interested in learning how we disseminate data.
|
|
minutes index
|
next
|