> > There's a much more compelling reason to accept FORTRAN as the 
> > language for BLAS and LAPACK: it's the language in which those two 
> > libraries were originally written and are maintained even today. A 
> > full rewrite would be a big task and would require
> extremely careful
> > testing.

> A full rewrite would be an interesting task
> 
> How many man years would be expected is hard to say
> 
> I guess it would take many years even if you had many highly 
> skilled people in both FORTRAN and the new language and by 
> the way what language would one want to use instead of 
> FORTRAN and why?

Given that there seems to be interest on the topic, I would like to add
that, as far as I know, FORTRAN stores array elements in column-major order
(http://sources.redhat.com/ml/gsl-discuss/2000/msg00425.html), whereas C
(just to mention one, but most APL implementation as well) stores array
elements in row-major order. My guess is that a complete rewrite would force
a change in the internal representation (or would have to pay the price of
cache misses) which would also mean a complete rewrite of all the interfaces
with other languages. In other words, it would force everybody to redo their
library bindings. Not a popular move...
-- 
WildHeart'2k6 - mailto:[EMAIL PROTECTED]
My digipics and blogs: http://spaces.msn.com/members/wildy2k5/

[[Philosophy 101: If a man speaks in the woods,  //
  and there's no woman there to hear him... is he still wrong?]]

----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to