Root Planning Meeting
Present: The ROOT Team, Marco, Andreas, Marcin, Peter H., Mohammad, David (R),
Peter (R)
ACTIONS:
o Danilo: Pass to Bill the cmsDriver command to dump a snapshot of the heap of
the application
Virtual Inheritance
o TStreamerInfo does not support for virtual inheritance, even if reflex and
cint dictionaries were.
Marco:
o About the PCMs: why not putting headers in the root files? This would make the
stop gap solution nearer to the final solution. It is not a blocker for LHCb
but it would be beneficial.
o Why cannot parsing on demand for functions work now? Because the infrastructure
for autoloading does not support functions.
o Let's see how parsing on demand behaves. Then we will see if we need a workaround
o Warning messages during rootmap reading: duplications (now detected). This will be
looked into and we can make the point afterwards, once it's clear if it's a fault
of LHCb or genreflex selecting too much.
o Unfortunately blinded by these warnings.
Peter:
o Head of repo frozen
o Will measure the memory usage again when header parsing on demand is validated
Marcin:
o Now working on the integration of the new Gaudi
o Using the LCG_CMT_68{_root6}
o Atlas is based on the CMT release. It is not clear if we can have an Athena building
against with the head of root.
Bill:
o Fix TClass::CanSplit a huge performance improvement
o Huge memory usage still
o Custom streamers are not properly used: Philippe working on it
Mohammad (FAIR-ROOT):
o Makefiles for VMC had to be changed to properly include c++11
o Rootmaps are used everywhere: need to use rootcling to generate them
instead of the present tool.
There are minutes attached to this event.
Show them.