Notes from CHEP03
 

Talks are posted at http://chep03.ucsd.edu.

Root Plans - Philippe Canal

• There is now a plugin manager, which keeps track of plugin handlers. This allows easy extension of base classes with new functionality.
• THStack allows using a list of histograms, eg to auto populate a divided canvas (finally!)
• TH1::Merge allows merging a list of histograms
• PROOF appears to be operational now. Parallel Root analysis, shown linear in number of CPU's to 64 nodes so far.
• can now generate SVG graphics files
•  The Win32 gui is now done, but slow. Next efforts are to speed it up.
  

Coming soon

• remove remaining limits on foreign classes
• better STL handling
• remove file size constraints (2 GB unix limit)
• TTree::Draw:
  • arbitrary C++ in Draw command
    
• New for gui
  • QT
  • gui builder
• online help in any Root process
• xml interchange - using xml as Root I/O alternative

Seal - Pere Mato

This is the LCG project to bring a common framework to LHC common tools

• object dictionary (so no need for rootcint dictionary)
  • use gcc_xml back end to spit out dictionary info
• object whiteboard (aka TDS)
• new C++ Minuit soon (March?)
  

Oval (CMS testing) David Chamont LLR

• sounds like our unit tests
• user interface is cvs-like. Config "xml-like". Perl
• comparisons include tolerance in xml file
• compare numbers not distributions
• easy for users/developers to run. In fact, not tied into any system running yet.
  
  
  

IGUANA

• using a G4 geometry overlap detector
  
  

ATLAS and CMS system tests

• NICOS is a very similar looking nightly build system, being run at BNL.  See http://www.usatlas.bnl.gov/computing/software/nicos/ for details.
• CMS' BOA seems to still be mostly vapourware and very buzzword compliant.
  
  

CDF Offline

• They established "offline shifts" to do builds, validation etc from recipes
• They established a help hotseat (a la SCS) to handle problems. They use JitterBug.
• They worry about NAN's, mostly in arrays. [I think vs.net traps single variables]
  
  
  

Random Notes

• Pere Mato says they will ensure that any migration from Gaudi to Seal is 'simple'.
  • They cannot afford for it to be otherwise either.
  • They have no nice linux IDE for enhancing/replacing gdb.
• Chatting with Rene Brun in the evening
  • ALICE is pursuing FLUKA as an option for its virtual MC
  • Rene is collecting geometries. We could send him ours. Ivanna H has a tool to convert G4 geometry (in memory) to Root.
  • I should plague Makoto/RichardM about G4 testing, notably how they missed the multiple scattering bug in 4.1.
  • ATLAS has apparently announced that they will not abandon CMT.
  • at least for now, it appears we are safe with Gaudi and CMT (confirmed by Pere Mato).
  • Rene uses ddd for debugging. He agrees shared libs are a pain.
• Can we get anything from McRunJob for configuration tools in the DPF?
• BABAR now doing distributed recon. Anything we can get from that?
• ATLAS Commander for production build
• AMI project for managing projects - Tag Collector
  
• Phenix using PostgreSql for database replication
• ALICE gui for G4 in Root (of course)
• Gaussian Sum Filter as alternative to Kalman? Tracy is aware of it.
• JACE for auto generation of JNI Java interfaces

R.Dubois Last Modified: 2003-03-31 20:16:07 -0800