Present: Joanne Bogart, Toby Burnett, Xin Chen, Richard Dubois, Dan Flath, Marco Frailis, Riccardo Giannitrapani, Berrie Giebels, Navid Golpayegani, Heather Kelly, Traudl Hansl-Kozanecka, Michael Kuss, Sean Robinson, Leon Rochester, Robert Schaefer, Alex Schlessinger, Tracy Usher, Karl Young
GlastRelease: Toby would like to make a tag now; we can make another one when the EM parser stuff is ready, probably by the end of the week.
(Alex) CVS trigger script is now in production. It runs every 5 minutes and does a build when there is something new to build. Another script for sending messages to mailing lists (one for GlastRelease and one for Science Tools) about the results of builds is just about ready; it will be activated in the next day or two. If everything builds correctly, a short message to that effect will be sent. Otherwise the message will describe the problems seen.
Alex is rearranging the directory structure for the builds somewhat, which could affect people who point to a particular build in their CMTPATH. He will send a message to softlist describing the change.
Navid will work on an enhancement which will make it easy to determine what has changed from one build to the next. He will concatenate diffs from the release.notes files of packages which have changed from one GlastRelease version to the next, then put a link to the concatenation in the releases table in the builds page.
(Toby) Still the case that Doxygen output corresponding to a given build is not publicly available. He has a student to help with project at UW to mirror SLAC builds there on the Windows system. Might be able to find a place there for the Doxygen output.
EM and data parsing: (Richard) Thanks to Xin's help, the parser for the V0 data format is now working at SLAC. Heather will be incorporating the changes needed for V1 this week.
The EM itself has had cabling problems. In the process of getting everything assembled, not always in the manner originally intended, 8 tracker cables broke. They are in the process of being fixed (and not a very pleasant fix, either). No data will be taken until next week.
The 200 Gbyte partition for EM data is ready to go.
Analysis GUI : (Richard) Jave + Swing + JNI look attractive, but the overhead (Java classes corresponding to many ROOT classes) would be high. Fermilab started on such an approach and gave up. Just now the Go4 project seems like the better candidate.
Insightful Miner: (Riccardo) The underlying statistical analysis functionality comes from a language called S+ (descendent of just plain S). One might be able to do something similar using Open Source tools, such as the R language. (Toby) One can extend Insightful Miner by invoking S+ functionality from it, but Insightful Miner is much more than a gui layer on S+.
Toby has a student working on a project, nearly ready, to read the XML description of Insightful Miner trees so that analysis of the data can then be done in the Gaudi environment. However Bill has since added new variables so some resyncing will be necessary.
Data processing pipeline: (Dan) Investigating a product called Opus.He has so far installed the server on a Linux system. He is working on switching it from using rsh to ssh. Java monitors can then be run anywhere.
It would need some customizing. For example, we want to keep track of what's in progress in a rdbms rather than using a flat file, and we use the SLAC batch farm for our processing, not quite what the designers of Opus envisioned. Source availability is an open question. Some customization is possible via a C library provided with Opus. We need to investigate whether this is sufficient, or whether we can get access to source if we should need it.
previous | minutes index | next |
J. Bogart Last Modified: 04-Aug-2004 15:40:55 -0700