Core Minutes 3/30/2010

• Collaboration Meeting presented plenty of opportunities for picture-taking. Here are some from Richard and more from Tom.

• Disks: (Richard) Now that Oracle has bought Sun disk prices have shot up: they won't give us the educational rate. The Computer Center is looking into alternate suppliers. There is an outfit in Berkeley which builds systems out of commodity disks which looks promising (but we have to verify that they work well enough).

• C & A at the Collaboration Meeting: (Leon) Pass 7 validation is ongoing. It's different from Pass 6 but not worse; maybe better. See details in talks . I spoke on Pass 8.

• Pass 7 plans: (Anders) See particularly the roadmap. We will reprocess with Pass 7. (Richard) We probably will redo the preceding 1-1/2 years of data. (John) So this will lead to a redelivery at FSSC? (Eric) Timeframe? (Richard) Yes; expect it will take 4-6 weeks. What about parallel ingest at FSSC? (John) We don't have it yet. It doesn't require much work, but can't be done until there are some naming changes at the other end. He will send email to the several interested parties.

• Pass 8: (Richard) There will be a Pass 8 workshop in Pisa in May. (Tracy) and we will have our next regular meeting this Thursday.

  (Leon) gave notice that the Pass 8 development team expects to make a lot of use of the event display. They need merit info in the WIRED display, and a more usable Fred on Linux. (Heather) is aware of the bits that are still needed for both of these.

• ScienceTools: (Jim) [who kindly supplied the following post-meeting.]

  • A new application, gtsrcprob, has been added to Likelihood. See JIRA LK-57.
  • I introduced a bug in gtselect when I implemented the default parameter functionality (using James Peachey's recent updates to hoops and ape). The time filtering parameters are read in as floats instead of doubles, so time selections can be off by several seconds for the tmin and tmax values. This should only impact GRB analyses seriously. A fix is in ST HEAD.
  • James Peachey updated the headers of the P6_V3* irfs to include the CALDB boundary keywords in all extensions. I've modified irfs/handoff_response to do this automatically now when these files are generated.
  • Toby is working on phi-dependence in gtexpcube.
  • There will be a new ST tag this week, v9r16p1.

• Unusual phenomenon: (Richard) Gamma ray emission was seen from a nova of V407 Cygnus, a symbiotic binary. People are quite excited about it.

• Passing the baton: (Richard) Jim will be taking Frederic Piron's slot as coordinator of the GRB group. Eric Charles will replace Anders for C & A.

• Documentation: (Chuck) sends the following:

  This past week was spent researching and working on the SRS Pipeline documentation. I'm currently testing Confluence 3.1 which, so far, looks like a more stable product as far as rich text authoring is concerned (at least in Firefox).

  (Richard) The Computer Center is now supporting Confluence, hence the move to a more current release. Richard is also happy with the newer Confluence; it plays more nicely with the Chrome browser on Mac. (Jim) would like to be able to run Chrome on his rhel5 machine, but it doesn't have all the required libraries. (Richard) suggests he make a request to the Computer Center.

• Scientific Computing Applications Department: (Richard) will become official on April 1st (really!). It will include the Data Handling and Geant4 groups, people from Atlas, xrootd support and so forth. We hope it will address (among other things) support for Fermi over the next 8 years.

• Core talk:

  • Externals

    • (Heather) The latest ROOT (5.26) has now been built for all platforms. There is a remaining issue with TMine, not expected to be intractable. (Jim) The next ScienceTools tag will use the new ROOT. (Anders) L1 will hold off using it until GR can catch up.
    • (Heather) Gaudi is not done yet, but she can see light.
    • (Heather, Tracy) There was a very fruitful meeting concerning obf involving Gregg, Jana, Tracy, Heather, Tony W. and, most crucially, JJ. His intent had always been to isolate the onboard filter from other flight software code but he was not able to take the time to do a thorough job. The outcome of the meeting was that Gregg, consulting with JJ and perhaps Tony W., will attempt to finish the job. When obj is broken free of most other flight software it will be much more portable. (Richard) How about on-the-fly selection of filter version? (Tracy) The filter has been very stable; it might not even be an issue. However if it does need to be addressed, it won't be easy.

  • Mac news (Heather) One of the macs now running leopard may be changed to snow leopard; we will then attempt SCons RM builds on that platform.

  • RM database clean-up (Heather) In the process of writing a script to check that all externals needed for a given build exist, Tony J. discovered that the database he consults for dependencies contains information even for builds which have been deleted. Kim will look into keeping the database in synch, but will proceed with care. Navid had already tried to do this, but his script would sometimes delete more entries than it should have. In the end he just cleaned up by hand on occasion.

  • (Heather) glast06 is sick in that it is impossible to log onto it as user glast. This has been the case for some time. It would be good to get it fixed, but we don't want to risk breaking RM builds.

  • SCons topics

    • Tom is starting work on a Command-line Installer for SCons which will complement Tony J's Gui Installer. No ETA yet.
    • He will continue to work on understanding why SCons installs don't always get built and why, even when they do, they don't always get registered in the database. Meanwhile he has done some patching of the database by hand and can continue to do so, at least for tagged releases, until the problem is understood and fixed. (Heather) suggests that if you can't get the install you want from the Installer you can try looking for it on u35.
    • Tom also fixed the triggerBuild program. It's supposed to be used to trigger a particular specified build but was failing to pick up one of the arguments correctly because of a typo. Then he added some database entries to make it possible to trigger LATEST, optimized builds (not a combination normally built by RM).
    • (Richard) LCLS uses SCons and is happy with it, something which came up in the context of the new Science Computing Applications department.

  • Non-SCons (Joanne) spent the last couple weeks otherwise occupied.

previous

minutes index

next

J. Bogart, Last Modified: 01-Jun-2010 15:45:29 -0700