Collaboration Communications

I looked into the CLHEP classes a while ago. 
The principle is quite interesting but there are big holes
in the functionality required. That is why, in the online, we
have decided to use commercial librairies such as those
from Rogue Wave: tools.h++  threads.h++ internet.h amongs other.
The problem with these is that they are quite expensive... 
Given enough time, I would vote to adopt the CLHEP approach:
develop our own classes so we do not depend on some other group
(CLHEP, or root) for the maintenance of our software. In practice
we do not have much time... 

This being said, I have the feeling that between the existing
CLHEP and the standard template libraries, one can probably
cover all the essential tools. I would propose, as I already did
to Torre a while ago, to "borrow" the CLHEP classes and create
our own version of them (call them the Star Foundation Class : SFC)
We would adapt, modify,upgrade as needed. We would also add new
functionality left out. There is a possible collision (or duplication
of effort) if Star Offline does decide to adopt Root. If we did
decide to adopt root, it would be wise to look at their definition
of the histo, ntuple, random generator classes. 
In addition to the CLHEP stuff, we will need classes for the 
creation, maintenance of link lists, vectors, arrays etc etc.
These are available under STL and could be easily extended in our
SFC. A word of caution, we have already experience (in the online)
that the stl are not really standard: different implementations
(from different vendors) have different behaviours...

My vote is to base our work on clhep but to separate us from them unless
we can share MOUs.

that's my 2(canadian) cents.

Claude

-- 
Claude A. Pruneau                    
Department of Physics and Astronomy, Wayne State University
666 West Hancock, Detroit, MI 48201 USA
Phone: 313-577-1813, FAX: 313-577-0711
Web Page: http://rhic.physics.wayne.edu/~pruneau