Core Minutes 8/25/2015

• ScienceTools: (Jim)

  • He sees problems running the tip test program. Joe, running nominally the same code at FSSC, does not [but see FSSC item below].
  • There have been crashes in the earth limb simulation. Keith has been contacted.
  • The healpix extlib was built against cfitsio v3.29, so it needs to be rebuilt using cfitsio v3.37.

  See Science Tools Development Notes for current activity.

• FSSC: (Joe) would like to grab the current set of externals from SLAC and check whether the cfitsio build he has been using is identical to the one at SLAC. If not, it could explain the different behavior of the tip test which he and Jim have seen. [Indeed, that is the case. Joe later reported to Jim that he had been using cfitsio 3.29 whereas SLAC is using 3.37 in ST-LATEST.]

  Current projects include preparing a patch release to fix problems seen with plotting on some platforms, and making an SL6 64-bit release which includes pointlike.

• Reprocessing: (Tom) No news.

• News from C&A: (Leon) Luca B. has created Confluence pages on a couple topics:

  • LAT at Launch + 7. What technical issues need to be resolved to enable Fermi to operate for more than 10 years?
  • Post-Pass 8 What else needs to be done after the first public release?

  There was also a presentation on Cal-only analysis. It would be nice to be able to use higher energy Cal-only events, but the ACD backsplash makes it at best very difficult. One thing holding up the works is the lack of adequate MC backgrounds.

• Core talk:

  • Gremlins in GLAST_EXT (Heather) Recently various things stopped working because of a missing python module. What was even stranger was that that module shouldn't have been there in the first place; it was not included in the original build.

    This kind of thing has happened before. There are a small number people with write access to GLAST_EXT. It's not likely anyone would accidentally install or modify compiled externals, but it can easily happen with python. She's looking for ways to protect against such accidents in the future, such as insisting that developers who need write access to the area get an additional account for that purpose, not used for regular development. (Jim) There might be some applicable strategies in this stackoverflow thread.

    (Heather) In any event, ST should move to version 2.7.6. GR has been using it for a while; somehow ST never got moved. She made a 2.7.6-gl1 version, including an additional requested package, and will try using it with ST.

  • rhel5 (Heather) has not heard any complaints about dropping rhel5. We won't be turning it off immediately, but she'll hold off on building new externals for it.

  • supersede not working (Joanne) heard from Franz Longo recently. He was having trouble building just a couple packages against GR 20-09-09, G4Generator and AnalysisNtuple. We found a way to get the build to succeed, but then he encountered run-time problems which seem to be due to undefined environment variables like FLUXSVCJOBOPTIONSPATH. I ran into similar problems recently when working with only facilities and calibUtil in a supersede area. I put it down to the fact that these are such low-level packages, but in Franz's case that's less likely to be the reason (though still possible; G4Generator is towards the bottom of the food chain). It should be investigated and at least understood if not fixed.

  • rhel6 builds not complete (Jim) Is this still an issue? (Joanne) nothing has changed since last report; there are another couple avenues to be explored. (Heather) It appears to be sporadic. Even when there are failures, we believe the builds are complete in all essentials; they're just missing one or more of the tarballs used for remote installation.

• Announcements (Richard) He and Tom will be attending the Fermi Collaboration Meeting next week. If you know of something which should be included in his Computing talk, get in touch with him.

previous

minutes index

next

J. Bogart, Last Modified: