Re: [ROOT] CINT and the GSL

From: Jacek M. Holeczek (holeczek@us.edu.pl)
Date: Sun Nov 17 2002 - 17:40:44 MET


Hi,
My 5 cents.
Some time ago (around mid 2001) I have also been looking at this problem.
If I'm not wrong then the main page for the GSL is :
	http://sources.redhat.com/gsl/
Today this page mentions a C++ wrapper (seems to be in "alpha" stage) for
some vector and matrix functionality.
There exists also a page with C++ wrappers to "special functions" :
	http://www.fnal.gov/docs/working-groups/fpcltf/Pkg/SpecialFunctions/doc/html/0SpecialFunctions.html
In that time, when I was working on this problem, I was solely interested
in a vector/matrix library in CINT/ROOT.
I developed an "approach" to this problem which was based on a very old
idea that the "g++" library was using - a "genclass" shell script which
generates the C++ link to GSL (generate classes from prototypes - like
"templates" which the CINT didn't like very much, at least in that
time). Such prototypes are then "parsed" by CINT/ROOTCINT.
I started with "implementing" a C++ based complex classes (float, double,
long double) and then I implemented the GSL "block" functionality
("gsl_block*" related).
Then I was ready to implement the "vector" and finally later the "matrix"
functionality, but ... unfortunately ... I realized that the amount of
new/delete operations in my computations would be much too high ... so I
needed to find another solution ... thus I stopped working on GSL.
(Fortunately, I had also realized that my needs were shut to 2,3 and 4
dimensional vectors and matrices, so I had found a C++ Tiny Vector and
Matrix Algebra package which I "genclass-ed" - this is possibly not the
most sophisticated way of working, but at least everything is
"inlined" by the compiler - and the code proved to be useful to me.)
Best regards,
Jacek.



This archive was generated by hypermail 2b29 : Sat Jan 04 2003 - 23:51:18 MET