Present: Joanne Bogart, Toby Burnett, Xin Chen, Richard Dubois, Navid Golpayegani, Traudl Hansl-Kozanecka, Michael Kuss, Julie McEnery, Sean Robinson, Leon Rochester, Alex Schlessinger, Karl Young
Data challenge planning: (Richard) organization, headed by the Analysis Group, is in progress; target for a plan is 1 May. So far, expect to produce one day's worth of data in September, including realistic background and some transient activity. By about November Science Tools should kick in though, with only one day of data, they won't be able to do much. Prerequisites include a geometry review, perhaps in June. The data challenge schedule also makes requirements on readiness of processing pipeline.
Toby and Theodore Hierat (first-year graduate student and possible future member of GLAST), are looking into producing an orbit's worth of data, a necessary step.
ROOT plans: Thanks to Ursula, several ROOT projects will get some needed attention:
It will take a little more work, in the form of writing a proposal, to get a second programmer to help with the ROOT analysis gui.
EM: The hardware has still got problems; the fix to connectors didn't hold. Additionally, half of the tracker channels are dead. However, Selim and Eduardo managed to produce an output ebf file, first with the mandated FITS wrapper, then (at our request) without.
Meanwhile with only minor changes Tracy and Joanne got Heather's code to run on Windows and Linux, resp. The test program, when run with the included test input ebf file, produces the expected output. However, the test program cannot read the new ebf file. Joanne, preferably with Selim's help, will look into this further today. [Post-meeting news from Tracy and Leon: they are able to read the file on Windows, although some of what's read is not reasonable. Joanne will retry, taking extra care to avoid pilot error.]
Release manager: (Alex) Messages are now being sent to the new mailing lists, GlastRelease-notify and SciTools-notify. Package owners are urged to include one or more author lines in their requirements file using the format of the following example:
author Joe Blow <joe@some.place>
so that email may be automatically sent to package owners when a package fails some step in the build process.
GlastRelease HEAD builds are already run off the trigger script but Latest is only built once over night. Soon latest tags will also be built by the trigger script as the new tags are made (however, there is some logic so that new tags of different packages occurring at approximately the same time will not each trigger a separate build).
Navid wrote a script which will do diffs on Changelog, making it easy to see what has happened from one tag to the next.
We're waiting on SCS to make the automatically-generated documentation Web-visible (in particular, the docs are being written to nfs space; we're asking that web servers on Glast01 and Glast02 be allowed through the firewall).
At some point, but not for a year or so, we'll have to come up with a strategy for deleting or archiving old builds.
Tag collector: (Navid) He has written up specs for an initial release and for later enhancements. The basic job of the tag collector is to automate the process of incorporating new tags into GlastRelease. ATLAS has a similar facility, but Navid believes it will be faster to write ours from scratch than to understand and adapt theirs.
Several of the proposed enhancements are similar or related to Release manager functions. Toby will draft the layout for an overall summary page.
Improve the view: (Richard) In order to make it easier to browse related packages, he proposes we (well, Navid) implement a CVSweb-like page, especially useful for container packages like GlastRelease. The page would only include those packages used by the given package.
Opus status: (for Dan) He has installed it on a Redhat 8 system on campus. It is close to working; he'll send some email questions to the proper people to get help with the last bits.
System tests: (Karl) Everything is now working off GlastRelease versions (rather than Gleam). Official switch to this new mode is imminent. Once it's done, information in the database pertaining to old tests will be flushed to avoid confusion and invalid comparisons between new old histograms.
Individuals can now run the system tests (or just a subset they specify) on demand.
Terminal server: (Toby) There was an unfortunate disk failure during the time of the switch to the new terminal server, requiring extra time to make the move, but it's all done now. One piece of remaining bad news is that rootcint is not running properly on the new server, even though the environment appears identical.
VCMT: (Toby) The latest version has a button to set the cmt path using a file similar in format to glastpack's CMTPATH file. The name of this file defaults CMTPATH, but a different name may be used.
previous | minutes index | next |
J. Bogart Last Modified: 01-Jun-2010 15:45:32 -0700