Core Minutes 3/15/2011EVO Future (Richard) investigated rumors that CERN is considering dropping EVO. Indeed, there are signs it could happen, though not immediately. They're seriously looking at an alternative, but it currently has no Linux client.
Meanwhile, there will soon be a native Android app. for EVO, followed quickly by one for iPhone.
ScienceTools: (Jim) No new tags since last time, but work is ongoing for an improvement requested by Jean; likely it will be tagged this week. In any case there will be a new ST tag this week as requested by FSSC.
FSSC: (Eric W.) was able to successfully build ST on Scientific Linux 6.0.
Snow leopard: (Richard) has exercised the Snow leopard build somewhat; not clear everything is working. (Heather) heard from one person whose only reported problem concerned a hard-coded path in the python external which ought to be fixable. Another user compiled ST from scratch on Snow leopard and has no complaints.
Catalog data: (Jim, Heather, Richard) Where should it go? extFiles is one possibility, but it could be inconvenient since updates require assistance from one of the few people with permission to write to the public GLAST_EXT area. CVS has the right privilege/access properties, but we're talking about a Gbyte or so, maybe too large.
Documentation: (Chuck) sends the following:
I'm still matching names to faces on the collaboration meeting pictures, and should have them ready to post to the web later today or sometime tomorrow morning. Then I plan to resume work on the Science Tools tutorials. I also received good feedback from the new members re: how they're using — or not using — the workbook. It's obvious that I need to update the four main "splash" pages, and I also need to do a better job of advertising the workbook's purpose, use, and availability. On other matters, Heather and Jim have given the go-ahead to start using kipac's ftools, so I'll be updating multiple links in the WB to their 'C' shell version as soon as Heather sends me the link that she wants promulgated.
Pass8 and related matters: (Leon) has been working on cosmic ray finding. The integration had been partly done; now it's more or less ready to go. The CR tracks are now in a separate collection, and the I/O and display and will work in the next GR.
TMine is being tested as engine for classification trees. There are still some discrepancies with the old method, to be corrected or justified.
Tracy is working on tracker filters.
The first version of the refactored WIRED is available. Many things which had been broken are fixed; additional improvements should follow quickly.
The main thrust for Pass8, now that the new pieces are all in place, will be to work on event analysis; e.g. the event-level analysis and then the classification trees. This will probably take the better part of a year.
Misc. sys (Richard) Daylight savings time wreaked some havoc with Nagios this weekend. The root problem was that u44 was full.
We're still thinking about how much new disk to buy. We have 55 Tbytes remaining and have asked the Computer Center about borrowing more.
Transfer from Lyon has gotten balky. Wilko is working on an alternative.
Power outages: (Tom G.) Likely dates for the three outages are
(Richard) The Computer Center will try to keep all server machines on our request list up, but might not find enough power cords in time for the first outage. Since it's shorter we could probably live with that, but have told them we do need everything up through the two longer outages. (Tom G.) That list is still tentative and assumes all xroot servers need to be up (35 machines). There are servers on the list which may not need to be up — it is difficult to get a definite answer.
cfitsio (Heather) James Peachey has emailed her that he's working on required patches to tip.
ft2util (Heather) with Jim's help has worked around fitsGen test app problem on Windows. So the way is now clear to add it along with evtUtils and embed_python to GR. Seems fine on RHEL4; if no problems turn up elsewhere will propagate to Pass7 and HEAD. Adding these packages makes it possible to build ft2util against GR only, with no need of ST. We're expecting a new version altogether of f2tutil in about a month. The new version will have fewer dependencies, always a cause for rejoicing.
G4 (Heather) is working on a build of 9.4 on Windows. She has run into a problem concerning the G4interfaces library which she expects to get sorted out which help from G4's SLAC hypernews list. Meanwhile, Franz has promised to put his G4Generator updates into CVS this week; mostly they are needed because of the new CLHEP namespace name.
TMine (Joanne) After years (2) of wandering, TMine has found a permanent home in CVS as the top-level module TMineExt-scons. In this incarnation it builds the ingredients for the external package TMineExt which GR is now using for classification (see above). We're ready to let SCons RM loose on it. Those who want to build and install locally can use scripts buildTMine.py and installTMine.py, documented in Confluence.
tagCollector.py (Joanne) made a small enhancement: if a container package such as TMineExt-scons has a top-level file called containerNotes.txt, it gets special treatment. The most recent version of the file will be included in a new HEAD build, no matter what the selection of subpackages to be included. The intent is that maintainers use this file as a place to keep notes about what went into a particular HEAD and why. The Confluence page on tagCollector has been updated. TmineExt-scons/containerNotes.txt provides an example of use.
RM builds (Tom S.) will turn off tiger builds [did so soon after the meeting] and is poised to start automated builds of TMineExt.
New windows box (Heather) reports that Kim is ready to start using it for CMT builds. Is there any reason to be wary? (Tom S.) doesn't think so. We should inquire to find out if there are limits on # jobs running concurrently on the new machine. The stringent limit in effect on the other Windows machines is not necessary for this one, with its 16 cores.
SCons meeting 10:30 AM PDT on Wednesday.
|
|
minutes index
|