Report for 2006-11-08

Topics I should cover:

• import code
  • Terry asked me to recalculate shared nodes with a smarter algorithm.
  • Wrote the code, unit tested the serial parts,
  • Code fails mysteriously on Terry's larger meshes.
• cxxtest framework proposal
  • unit testing and charm++: test to see if functions match their pre/post conditions.
  • incorporated into my local ParFUM? , perhaps it would be useful in the source code as a whole?
• CPSD progress
  • Learning the system
  • writing test cases for the parts I don't understand in order to verify my knowledge.
  • Code appears to "work" adaptively for the 1vp case, waiting on word from TAMs people for final confirmation of this.
• system administration
  • updated the TWiki installation on charm
  • set up trac repositories and wikis for the CPSD group, already proving it's worth.
  • need to eventually move cpsd over to our svn servers. I'll move trac at the same time.
• multi-dimensional arrays.
  • Began with innocent question: why can't we create construct dense charm arrays with the constructor? We can do for the 1D case.
• Tutorial
  • tutorial is in need of SERIOUS revisioning
  • Charm workshop presentations much better at teaching the fundamental concepts.

Future work

• future work
  • not here!