Core Minutes 3/2/2010

• ScienceTools: No one was present to give the report, but you can see it in Confluence. Main news is the new ScienceTools tag, v9r16p0.

• FSSC: (Eric W.) The ingest of the new tag is going very smoothly. The work already done, merging patches from FSSC into the SLAC repository, has paid off.

• Documentation: (Chuck) sends the following:

  Priorities in WB World continue to be the generic pipeline, which is taking a substantial step forward thanks to input from Tom, and Pointlike, which I've not worked on recently. The gtgrb tutorial is still at the top of the update list, and this will be done as soon the soon as I can run the GRBanalysis script for an SCons build without problems.

• GR Pass 7: (Anders) No news at this point. Eric C. is continuing to work on it. Expect a report in Paris.

• GR Pass 8: (Tracy) Everything is wonderful.

• Disk problems: (Tom G.) sulky55, which serves user disks, continues to have problems when hammered upon. Today there will be an outage to back out a suspect change to a network card. There is also talk of new hardware but no definite plans yet.

• Core talk:

  • rhel4opt builds resurrected (Heather) There actually never was a problem with "regular" rhel4opt builds, only with manually triggered ones, which go through a somewhat separate path. The default platform in the script used to trigger builds was rhel3. Kim has now updated this script and its documentation.

  • Over-enthusiastic cleaning (Heather) Because of a miscommunication all of our Beamtest releases were deleted. Michael Kuss was able to provide some old Installer releases and we're now rebuilding some of the later tags. We also need to repopulate the RM database and will archive releases (in addition to keeping them around on disk).

  • SCons matters

    • (Toby) is concerned that recent updates, including Joanne's latest changes for SCons builds, have not made it into the new ScienceTools release. (Joanne) None of her changes affect FSSC builds, so in that sense it's not a problem. (Toby) However documention in the workbook should refer to a tagged release, and ST tags are getting to be few and far between: the last one prior to v9r16 was in early December. (Heather) Jim was anxious to get this tag out before leaving for Hawaii, but he will probably be willing to make another tag soon for the new version of ROOT. We should coordinate so that other changes are added at this time.

    • (Joanne) implemented a somewhat neater solution, in a tool called addLinkDeps for bundling up platform-dependent environment set-up for building a (not dynamically loadable) shared library. It handles other types of targets as well, but so far there is no platform dependence in those cases, so SConscripts have only been updated to use it for the one case.

      The TRAP_FPE macro was being defined too widely (should be Linux only); that has been fixed. And she has put in a work-around for a long-standing problem on Windows with building shareables with SCons. The tool we're using for exporting symbols (bindexplib.exe, part of the ROOT distribution) behaves differently — and, for us, wrongly — when processing modules including _declspec declarations. The work-around is to surround such declarations with #ifndef SCons ..#endif. Only hoops and st_facilities had to be touched.

    • Gaudi (Heather) The new version of Gaudi has removed several lower-level services we're dependent on and replaced them with things having different interfaces. Most of these changes occurred already in v19. Currently she's working on modernizing GlastRandomSvc to use the new interface.

    • SCons report from Tom He's looking into revamping the flaky protocol now used to generate CMT-style tags from SCons-style tags, and into writing a program to handle the export-symbols function on Windows, so that we won't be dependent on an external tool (bindexplib) which doesn't always do what we need.

previous

minutes index

next

J. Bogart, Last Modified: 01-Jun-2010 15:48:12 -0700