Core Minutes 10/11/2011

• ScienceTools: (Jim) There is a new ST tag since last week: 09-25-01. And there has been a new tag of st_facilities to fix a minor problem. See Science Tools Development Notes for more information.

• FSSC: (Eric W.) got help from Jim in migrating to new gtsrcmaps. They made a new tag on Friday and are doing internal testing. It appears to resolve all previously-reported memory problems.

• Pass7: to reprocess or not to reprocess: It's still up in the air. (Leon) By some measures (e.g., better cal track doca, cal track angle, ctbcore) there is a lot to be gained. (Toby) I'm not convinced that there is a science case.

• GR/Pass8 (Tracy) Studies are in progress; nothing new to report yet.

• Core talk:

  • Gaudi (Heather) There was a note in the Gaudi talk mailing list of plans to discontinue Windows support at the end of the year. It may not be that much of a blow to us. They only officially support VS 2003; we had to make a VS 2008 build ourselves. Possibly more worrisome: even the most current release of Gaudi officially only supports gcc 3.4. but she's hopeful that it won't be too difficult to migrate the relatively small part of Gaudi we use to gcc 4.x.

  • obf (Joanne) is still in conversation with JJ about a couple include files, but there is a usable version of B3-0-0 installed in SLAC GLAST_EXT for redhat4 32-bit; comparable installations could easily be made for all other supported Linux platforms. OnboardFilter code which will run with either obf version is committed and tagged, as well as upgraded SConsFiles and IExternal/obf required for the new code. GlastRelease is still using B1-1-3. So far B3-0-0 is only formally available for Linux.

  • GR and redhat 5 (Joanne) took an inventory of redhat 5 externals. We're not that far from a complete set for GR. (Heather) recently built OmniOrb and fox. (Joanne) will take care of TMine and obf. The only remaining one is gaudi.

  • ST and redhat 6 (Heather) had asked Kim to build all the externals ST needs on redhat 6; he is now nearly done. Given the trouble-free FSSC experience, we're hopeful that compilation of the ST code with the newer gcc compiler will go smoothly.

  • redhat 6 32-bit? (Heather) sent out a query: do we need to support it? Whether or not it would take extra hardware, it would take extra work, such as building and maintaining another set of externals. From responses so far, there seems to be no compelling reason for it. For those out in internet land: if you think you need it, now is the time to speak up!

  • SCons Windows RM (Tom S.) The Windows RM fell on its face again for no discernible reason. The Qt config file was still there; the registry appeared to be untouched (which shouldn't matter anyway, since we switched to using a file). lsf is not the problem because the jobs are starting, but there are lots of "can't connect to database" errors. (Heather) Is it possible to make the build by hand, just so that it will be available to the Installer? (Tom) Yes; he can run the various RM programs from a console window. [And later on Tuesday he did and reported as follows:]

    I'm going through the process of running all the SCons RM programs by hand on glast-win04 for the debug version of the release tag and everything seems to be going fine (it's building the distribution packages right now).

    After I started the compile step, just for fun I deleted the Optimized windows build to see what would happen. It just picked up and started the process without any errors and that thread is now on the testing stage.

    I didn't do anything to the configuration setup except resave the configuration file. I don't know if that had anything to do with it or just logging in and out of the glast account on glast-win04. It's still a mystery as to why exactly it stopped working but it is working now.

  • CVS checkout glitch (Heather) There was a period when RM checkouts of GlastRelease were quitting part-way through, though not always at the same spot. But the problem has mysteriously fixed itself.

  • L1 and SCons (Heather) made a new SCons GR build on the L1 branch, 17-35-24-gr08, which fixes the problem in G4Generator which had caused Systests to run very slowly. (Leon) Systests are already done for that release.

• Announcements

  • SCons meeting this week at 10:30 on Wednesday.

• AOB

  • (Joanne) tagCollector as currently implemented makes several checks to be sure that arguments are sensible, but does not do any validation of the input file which specifies which tags are to be changed. It will make a partial, useless new HEAD tag even if packages or tags specified in that input file don't exist. Jim has requested that, in such cases, tagCollector abort before starting to tag. I agree this is the right behavior. It's on my list of things to do (but not at the top).

  • VMs for legacy systems (Joanne) spoke to Steffen Luitz and Homer Neal about the BaBar LTDA (Long Term Data Access) architecture (see Steffen's presentation from the March 18th PPA SCA meeting, especially slide 12) to understand how applicable — or not — it might be to Fermi and why certain choices were made. I doubt such a self-contained system is feasible for us, especially while we're still taking data. Some components (notably GlastRelease) may need to be frozen long before others need to be, or even can be. I was pleasantly surprised to learn that Steffen didn't think such a hybrid architecture was out of the question; on the contrary, he believes that is likely the wave of the future.

previous

minutes index

next

J. Bogart, Last Modified: