ScienceTools: (Jim) No report this week.
(Toby) has made some structural changes to pointlike and sourcelike for smoother operation with SCons while maintaining the ability to use CMT (but he really would prefer we get back to a one-build-system regime):
FSSC: (Eric) John uncovered problems with Mac platform which Eric is now attacking. After that he'll work on a true snow leopard port from sources (the earlier claim that tiger builds could run on snow leopard isn't quite true). (John) has been occupied with the db server: "some hardware issues and minor things like that".
Documentation: (Chuck) sends the following:
Most of this past week has been spent working on the Google Group Fermi LAT emergency backup sites, which are now online. Postings to the Fermi LAT News group are public and may be viewed at http://groups.google.com/group/fermi-lat-news?hl=en.
There is also a Fermi LAT restricted site (i.e., password protected) containing primarily static information to be used by management and those who will need to access this information in an emergency such as the recent power outage. To access this site requires that you have a Google Gmail account. It should be noted that you can then use that account login name and password (or that of an email account that you have designated as "associated" with your Google account). Access to this group is by invitation only and membership in this group is not yet decided.
Next up on the priority list is to post Joanne's updates to the SCons pages [(Joanne) in progress; about 2/3 done] to the website, prepare for the next week's newsletter, and advance the cause on Pointlike.
GR:
Pass 8 (Leon) Plotted chi-square of track versus angle for high-energy electrons. There is a very strong, essentially linear dependence with large errors at large angles. Then he looked at muons. Even there the effect is visible. (Reminder: originally the slope-dependent hit errors were "measured" in MC and fit to a model. It looks like this procedure may have to be revisited.) This effect washes out subtle alignment corrections.
Tracy has tried out a kludge correction: he increases the calculated error by a correction factor which is proportional to the cluster width, with the constant adjusted to flatten the muon plot, but it only reduces the the problem for electrons by ~15%. So there some actual physics going on here... we're thinking delta rays...
(Tracy) New tags in the v18 and v17 series (v18v1r1 and v17r35p3, resp.) have a new Overlay package which addresses the Calorimeter overlay energy problem. (Leon) The fix works above 100 Mev. Below, we're still losing some energy.
Pass 7 (Anders) No comment beyond Philippe's latest results which can be found in Confluence. There are problems with effective area calculations, front and back.
(Liz) The only changes in systests for the new v17 tag are in the overlay test, as should be the case. She doesn't yet know whether the differences she sees are the expected ones.
People: Congratulation to Kim on the birth of son Derek! Welcome to Tom S(tephens), now working for us 100%. And Wonder Woman Heather is back from maternity leave. Many reasons for jubilation!
RHEL3 builds have ceased as planned. It went smoothly however there was one unexpected side effect. No one remembered that the per-release branches were being automatically created along with RHEL3 builds, so they have ceased as well. It's convenient for Heather to have them, but they also make a certain amount of clutter in CVS tag land. The plan had been to stop making them when we move to SCons. Heather will ask Navid if there is a reason other than clutter for turning them off.
Externals on the move (Heather)
Root Root 5.20.00-gl6 has a fix specific to Windows, something Eric C. needs for TMine on Windows. Heather would like to make a new GR HEAD with it. We might consider going to a newer Root altogether, but in that case it should be compatible with the new Gaudi which will be needed for Windows VC90 builds.
ape James Peachey wants to update ape to 2.6
PythonTom G. has requested the ALPpy add-on for Python, which requires a couple other new packages as well as newer versions of some we already have.
(Toby) suggests that it might be a good opportunity to move to Python 2.6 while we're at it. (Tom) has exercised it some but has not yet attempted to use it in production. (Joanne) In principle SCons would support it (definitely would not support 3.0), but their testing is primarily with python 2.5.x or less. We would have to do our own verification with 2.6 to make sure everything we care about works on all supported platforms.
ASP and SCons (Jim) has not had a chance to do anything more to understand the problem seen concerning exception classes. This problem is not specific to ASP; it could potentially affect ScienceTools as well. (Joanne) is willing to look into it. Any fix will most likely affect SConstruct or other SCons machinery files.
GR and SCons (Joanne) Job options plan is partially in place:
New SCons versions (Joanne) continues to spin her wheels trying to get the latest checkpoint release to behave properly on Windows platforms with both 7.1 and 9.0 installed. See last week's report.
previous | minutes index | next |