ScienceTools: Jim presented the report. He also mentioned problems people are having with binary releases of Science Tools. See this email thread.
FSSC Report: (Eric W.) Working on porting v9r11.
Documentation: Chuck provided this report:
Helen has posted some of the Hiroshima pictures to the web, namely the 'People' and the 'Japanese Cuisine' shots; Miyajima and other location shots will follow, presumably later this week. You can view them in the Picture Gallery, which is linked from the SAS 'splash' page. Meanwhile, I did make a new tag for the workbook; we're now at v6r10, and Berrie has updated the mirror site. Mark Strickman submitted the only update since, and that was to include the information about downloading the 'known_hosts' file when installing SSH for Unix and are having problems between a SLAC machine and a non-SLAC machine.I'm also making some progress on the pyLikelihood references, but have compiled a relatively short list of questions that I need answers to before I can proceed. Hope to get those answered later today.
GR tag (Heather) Latest is v17r17. It includes upgrades to EventIntegrity, HepRepSvc and G4HadronSim [and probably others I didn't catch. ed.]
Externals path (Heather) Some time ago the decision was made to add a component to external library paths to indicate compiler. This is primarily to benefit Windows users who might have multiple versions of Studio installed. It has been implemented fully for externals used by SCons, but is independent of the move to SCons. Some of the externals used by CMT builds also have the new structure. Most recently the change was made for G4, affecting both Geant4 and Geant4Runtime. For backwards compatibility a new path variable was added. However all this also affected G4HadronSim since it's in charge of finding tables for physics lists. Runs were made in the in-between state where G4hadronSim had not been updated to use the right variable and hence could not find the tables. (Leon) It muddled on without them and it probably didn't make much difference in this case, but.. The tables are just ascii files; there is no need for them to be associated with a particular compiler version. And if we're routinely using an alternate physics list, maybe it shouldn't be "alternate".
(Heather) Would consider reverting the change to external library paths; it's causing more trouble than had been anticipated. (Navid) At this point, that would be non-trivial, particularly for SCons. (Consensus) Probably best to stick with it.
Crawler (Heather) Tony has implemented the required fixes to handle the new merit header, which now includes an int to specify version of the ntuple. He plans to re-crawl run that have the new header so that they will no longer be marked as bad. Tony has requested to be more in the loop concerning GR updates. Heather has offered to make monthly appearances at the Data Handling meetings to attempt to provide such information.
Version string(Heather) There is a string stored in our MC jobs, that is set via the JobOptions and is used to specify the GR tag used for the job. Tony reported trouble with recent MC runs, where the jobinfo string appears to reversed the version and revision numbers of the GR in question. As this quantity is set by hand in the JO files, and is usually just copied when performing more jobs, it can easily be incorrectly set. At the present time, Heather believes L1proc does not set the jobinfo string at all in its JO files, and thus this is set to the empty string. We now have the ability to store the results of "cmt show uses" in the header of our output files (though admittedly, there are times when the cmt show uses call fails). It may make sense to abandon the jobinfo string in favor of a more automated approach of determining what tag was used for a particular job. We should discuss this with Tony, as well as Toby who created this jobinfo string to begin with, before making changes.
Archiving status (Tom) We're good! 5 4.6 TByte disks have been backed up, checked and reformatted. The remaining data to be archived comprises directory trees of a similar nature — nothing more recent than launch; much of it much older — but no more complete disks. There was one instance, the first encountered in the whole endeavor, of data on u25 which does not archive correctly after repeated attempts. Everything else is ready to go.
Python 2.6 (Emmanuel) He was able to build it for RHEL4 except that a particular module failed. Maybe this module isn't in 2.5.1 (Heather) We might try doing without it.
ACD updates (Heather) Still waiting for action from Eric C. The new stuff is not in CVS.
SCons RM (Navid) Mac's are still down because of a problem with AFS tokens.
Windows is back up. He restored some necessary register values which had mysteriously disappeared. The Windows builds mostly worked, however what failures there were were not detected by RM because of the different end-of-line characters for Windows. He's working on fixing this.
He has requested a new AFS volume (probably will be available shortly) to hold externals for (Mac OS version) Leopard.
ROOT on Mac (Heather) It turns out some special compile flags are needed to build ROOT, pyROOT on the Mac.
CHS (Navid) The RM build was incomplete because some packages had no SConscript file at all or one that was of an obsolete form; others had the file but had not been tagged. These problems should be straightforward to fix.
SCons/Windows and GoGui (Joanne) spent the (much) better part of last week at Yosemite. Upon returning, she cobbled together the rest of the machinery needed to incorporate dependencies in solution files so that projects will be built in the correct order. See the Windows log for some evidence.
Fred (Emmanuel) Made a complete circle, but learned something along the way. In particular, he's able to show that the Corba connection is not the problem. However, he still sees this error when running Gleam with Fred:
FXRbToolBar::create: trying to create window before creating parent window.
previous | minutes index | next |