Core Minutes 3/1/2011

• ScienceTools: (Jim) Since the last meeting a new tag of tip which made use of a TEventList in reading ROOT files had to be backed out because it had a bug, one which would not be easy to fix. We reverted to the old version which uses a temporary file. A new tag of fitsGen uses a ROOT TEventList directly, thus avoiding the tip problem. Masa has made several improvements to pulsar tools. See Science Tools Development Notes for details

  In other news, Jim used the SCons tag collector script to make a new HEAD build. It appears to have worked correctly.

• FSSC: (Eric W.) It's very quiet here. The site is back to normal after the security problems. (Tom G.) Is FSSC ready for P6_V11 reprocessed data? (Tom S.) Basically, yes. He would like to take a look to see what's differenct in the new data. (Tom G.) Very little. (Tom S.) is at Goddard this week to help train, the new hire, Changxue Deng, who will be helping with the database among other things.

• Pipeline: (Tom G.) We're still runnning at P6_V3; have not yet upgraded to P6_V11. (Richard) We're aiming for a full reprocessing with Pass7 approximately July - October. We don't yet have a place to put the Pbyte or so of recon files.

• Documentation: (Chuck) sends the following:

  So far, I've updated all of the ScienceTools reference pages, though I'm sure that I'll have to revisit gtexpcube2. In the meantime, I've started on the tutorials and barring any surprises, I should finish updating the "Explore LAT Data" tutorial later today. So far, I have been experiencing problems with our current versions of the ds9 and fv ftools, which we link to from numerous places in the WB when running on SLAC Central Linux. So far, I've been able to work around the problems by linking to Kipac's tools (source /afs/slac/g/ki/software/etc/profile.xray) when I have to run ds9 or fv, but it's cumbersome at best. However, Heather said that she would look into linking in kipac's ds9 and fv to our astrotools, which looks like it might solve the problem — providing there are no other obsolete versions in our astrotools. Prior to next week's collaboration meeting, I need to clean up the WB files that have been "touched" while making the recent updates and make sure all files are checked into CVS. I can then cut a new tag so Berrie can update the mirror site.

• Pass8: (Leon) The Pisa meeting was very productive. We are on track, or nearly, with the work schedule from the first Face-to-face. We now have a tentative plan for the next 6 months. We would like to have another meeting in that period, perhaps in Santa Cruz if travel budgets allow.

• G4: (Heather) We are getting close to moving to version 9.4. (Johan) Franz was able to compile GRv18r8p1 with GEANT4 9.4 and CLHEP 2.1, at SLAC. I'll run some tests in the next few days.

• Power outages: (Tom G.) There will be a couple power outages at SLAC in April, each possibly lasting as long as 48 hours. We have compiled a list of servers of interest. There are about 30 xroot servers; currently they can only be turned on or off as a group. We are investigating ways to change this so that some can be off and others on. There are a similar number of other miscellaneous servers, herein enumerated.

• The French connection: (Tom G.) There is an ongoing problem with glastlnx22 and its handling of incoming xroot files from Lyon. The problem seems to be an excessive number of connections between the Lyon client and the FUSE file system. Work to solve the problem continues with an expectation that a newly repaired xroot proxy server will supplant the FUSE system. Timescale for tests of the new system: 1-2 weeks.

• Core talk:

  • cfitsio (Heather) Move to new version is on hold until James Peachey has a chance to make required patches to tip.

  • ROOT (Heather) made a new build of our current version of ROOT called v5.26.00a-gl4. It's only strictly necessary for Windows, vc71 compiler, since the old build for this platform was defective.

  • TMine (Heather, Joanne, Leon) is in a state of flux. It's not quite working with Gleam on Windows. Some problems have been sorted out but possibly not all, in a couple new versions. The rapid evolution causes problems of its own; we're not in the habit of installing new versions of an external twice a week or more. [Update from Leon this afternoon: "Okay, I got all my ducks lined up and it seems to be working now! Thanks everyone! Onward to the comparison..."]

  • Windows RM builds (Heather) are in much better shape than they have been in for some time: they're basically working for both CMT and SCons RM. There are still some lesser issues to be dealt with:

    • It's not possible to fetch Windows source zip files for CMT builds using the link on the RM pages. However, the files do exist on u05, accessible via ftp. [update: appears Kim has already fixed this]
    • identity and known_host sporadically disappear and we don't have a way to correlate with other activity. Joanne has suggested they could be monitored so at least we would know more precisely when they disappear.

  • New windows box (Tom S.) SCons RM is successfully using glast-win04 to make Windows builds. Whatever failures there are are not unique to this environment. So far it is only making vc90 builds because we haven't asked it to do anything else. It is not able to send emails, but there is no need to do it there; it can be done just as well from Linux. [In recent email Kim said he has tested all CMT build steps manually on the new box and they're all working. He is about ready to try automatic builds.]

  • Unit test infinite loops (Heather) Kim has spotted a couple in CMT builds. One in reconRootData has been fixed. Another in AdfReader is more mysterious since the package has been stable for months.

  • Snow Leopard (Tom S.) It's working! We haven't even had any recent instances of afs token loss. (Heather) meanwhile tiger builds have become a bit flaky, but since they're going to be shut off on March 15th it's not worth the effort to investigate.

  • Red dot display (Heather) is still lacking for SCons builds. [Kim writes that Karen has a draft page; work is still in progress.]

  • Misc. (Joanne)

    • root maps and G4Generator: Heather had successfully created a root map for G4Generator from a CMT Windows build of the library. We were unable to do it for the SCons-built library for either compiler (vc71, vc90). However, there were some possibly-significant differences in the way the library was being built with SCons (missing a couple G4 dependencies; missing some compiler options). After fixing these and rebuilding the G4 library Joanne was able to make a root map for the vc71 version. On vc90 it still fails. But we are, we hope, on the verge of moving to the new G4, so further debugging efforts are on hold until then.
    • New tag collector script: I've made a small change since Jim used it, affecting only the form of the non-essential file packageList.txt associated with each container. That file is there to provide an easily accessible record of which package versions belong to which container release (the same information is in the RM database). The upgrade to the tag collector script fixes a bug which was causing extra blank lines to be inserted.

  • SCons meeting None until the week after the Collaboration Meeting

previous

minutes index

next

J. Bogart, Last Modified: