Core Minutes 04/17/2002

Present:  Riccardo Giannitrapani, Heather Kelly, Michael Kuss, Sean Robinson, Leon Rochester, Alex Schlessinger, Tracy Usher, Karl Young

• Build Tools
  Toby, Karl, Alex and Heather had a VRVS meeting on Tuesday to discuss the future plans for build tools.  Here is a summary of the meeting:

  • Alex has created a test version of a script called glastpack.  It is available from:
    $GLASTROOT/ground/scripts/glastpack.pl
    This script accurately performs a recursive checkout using CMT and the CMTPATH.  The script will also retrieve the most recent tag that matches a wild card in a use statement.  It is expected that Alex is working on the documentation for the script and will provide it in the next couple of days.

  • Toby introduced the notion of test_releases.  Currently, our releases are expected to compile and run on both platforms.  A test_releases is actually an iterative process - where patches are applied until the code works on both platforms.  A developer would tag a package, such as Gleam.  Then an automatic build tool will test the new package on both Windows and Linux - making sure it compiles.  If a package fails, the package owner is notified via e-mail.  Any required patches are applied, when the test_release compiles and runs on both platforms - a tag is declared as a release.  Karl and Alex will be writing up a proposal for supporting test_releases, which should be ready in a week.

  • Toby provided Karl an example script that extracts error messages produced by gmake - as Karl feared it uses an error library to accurately describe the error.  Karl will bite the bullet and will create an error library for our use.

  • It appears that private use statements are not processed properly by CMT on Linux - libraries of packages that are used privately are not compiled as needed.  It is recommended that developers cease using private use statements in their requirements files.  In addition, developers should not leave the version tag for a use statement completely blank, rather, wildcarding is encouraged, for example:
    use ROOT v3r2p*
    rather than:
    use ROOT

  • After the meeting, Karl and Alex spoke to Richard about the build tool plans.  They have started to test the feasibility of storing all packages in a flat directory structure.  Toby had discussed the possibility of storing all packages in the release area in a flat directory structure - this would eliminate the need to store duplicate copies of the same version of packages.  Alex and Karl will evaluate the requirements this will impose upon the build tools.

• Coding Rules Checker
  http://www.fisica.uniud.it/~riccardo/research/glast/rules/rules.html
  Riccardo has written a coding rules checker as he prepares for the G4Generator Code Review.  The scripts are written in Ruby.  The scripts provide an automatic mechanism for checking adherence to our coding rules.  Riccardo has started work on a GUI.  He welcomes comments.

• MC ROOT
  Almost there.. about to import a new package named RootIo.  This will replace our old glastRootIo.  Why a new package?  This provides a clean break with the old glastRootIo which uses the old ROOT classes.  In addition, the new package adhere to our package naming conventions.

• Tkr Update
  Leon reported that there are some updates being made to the Tkr digi classes.  The modifications will impact 3 packages.  The next step will be to implement the ROOT class for Tkr digis.  Heather will provide some additional details about writing ROOT classes. 
  Tracy stated that there are discussions taking place between Leon, Tracy, and Bill about the future directions for the reconstruction classes.  There are plans to modify the track fitting classes to separate the algorithms from the data.  Implementation details will likely be worked out with Toby.
  Riccardo reported that Claudia informed him that they are now up and running on Windows and will commit their Bari TKR digi algorithm in the next couple of days.

• G4Generator Code Review
  Next week, during the regularly schedule core meeting, we will have the G4Generator Code Review.  The code will be tagged this weekend.  This will allow the reviewers to go over the code before the actual review :)

Action Items

• Alex will complete the documentation for the glastPack script.
• Alex and Karl will write up a proposal for test_releases in the coming week.
• Heather will send Leon and Tracy additional information on writing ROOT classes.
• Riccardo will continue to prepare for the G4Generator Code Review.

H. Kelly Last Modified: 01-Jun-2010 15:45:13 -0700