RM backend
There was a problem with the unit tests where the shared libraries of the externals were not being found. This turned out to be a bug in the RM backend which is now fixed for rhel3, rhel4 and 64 bit builds, rhel5 is on hold pending installation of the mysql libraries. Navid has contacted SLAC unix-admin.
Windows builds have been turned off due to problems that need to be addressed. Fixing those issues is on the todo list.
64 bit build unit tests are causing LSF machines to crash. John V reported that some of the unit tests were gobbling memory. Richard stated that the LSF machines will reboot due to excessive memory use. Navid will turn off the 64 bit builds for now and will contact Eric and John to get a list of affected unit tests. Developers will be notified in the hopes that those unit tests can be patched.
Navid will start up the rhel3 builds, they should already be set to go.
g2c versus f2c
Navid summarized the result of discussions in the software meeting and via email. On rhel5, according to the SLAC admins, libg2c is not provided. We can eliminate its use in GR, and hope to do the same in ST. For now we will have to distribute the library.
Navid has committed the fixes necessary to allow an overriding directory in ST. Navid has tested this out in the simple case, and Joanne will continue trying it out. Once Joanne gives the ok, Navid will commit the changes for GR and CHS as well.
RM frontend
Karen has uploaded the new RM to CVS. Tony will put the new RM into glast-ground sometime this week. In the meantime, Karen asks that we take a look at what is on tomcat03, and provide feedback.
Navid has released RMviewer 0.3 which is now multi-threaded. Next up is to improve the speed of the application by ditching the RM libraries and write RMviewer specific code to connect to MySQL. After that, Navid can add in the actual tag collecting functionality.
GoGui
Alpha version 0.8.8 is now available.
Excluding packages is now all working. Explicitly excluded packages are crossed out in the gui, which non-explicitly excluded packages are grayed out. Users are prevented from building excluded packages.
Packages are marked with T (tag), H (head) similiar to what we have in MRvcmt.
Emmanuel has been signed up as a GoGui user and has reported trouble running GoGui on a mac via an xterm. After further discussion and googling by Joanne, it seems there is a known issue associated with running some applications via an xterm.
ScienceTool builds
Navid has updated the pointlike and skymap *.i files (which are the SCons equivalent of the *.in files for python). The updates have been tagged v9r8p2. Unfortunately, the package owner (Toby) is not maintaining the *.i files, so each time there is an update to the *.in file - Navid has to make an update. In fact, there are further changes for ST v9r8p3 that Navid has to implement.
GR builds
Emmanuel has successfully performed a global build on GR v15r49 and is now working committing the changes to the SConscript files. He is hoping to have them all committed and tagged by this afternoon. Next up, Emmanuel plans to continue experimenting with GoGui.
SCons
The bug fix for handling recursive copying of the include files has been implemented. There are other Bryson updates for CHS that Navid plans to migrate over into GR and ST, such as the option for package owners to define environment variables in wrapper scripts. Also OBF has multiple include paths, and Bryson had a workaround for that as well.
Transitioning JO files to the top-level directory is on hold pending completion of other higher priority items.
Navid has not yet contacted Tony about the new installer. First Navid wants to determine the mechanism for creating the tgz files that the new installer will be delivering. While RM could be made to create those files, as was done for CMT, there is interest in having this step handled through SCons. This would allow developers to create their own tgz files. This feature has been desired in CMT-land. Richard also had a question concerning the use of a single tarball for each build, rather than providing separate files for each package. Navid confirmed that this is his intention, where the mechanism to continue with the download in case of interruption needs to be available. Joanne (and Heather) expressed some concern that downloading a single large file may be impractical. Navid believes that a single file would actually be less prone to download troubles that many small files, due to the elimination of numerous file download initializations. Navid will try to gather some statistics to demonstrate.