(j3.2006) decimal floats

[Van Snyder]

On Fri, 2008-01-11 at 16:54 -0600, Bill Long wrote:
> Any implementation that actually has to implement a new real 
> kind (any kind) faces an enormous task of writing new versions for all
> the intrinsics that take real arguments.  For intrinsics like 
> dot_product, multiple new versions. Ugh. 

Hopefully, you can generate many of the intrinsics, such as dot_product,
maxloc... from some kind of a template.  BesselJ0 might be a different
story, since one needs different approximations for different accuracy
requirements (assuming you want maximum efficiency for each accuracy
range).  We've generated all four versions (single,double)X(F77,C) of
each component of our 300+ component math library from a single version
for more than 20 years.  With a staff of only three, you have to
automate.