[J3] More Bessel functions

[Dan Nagle]

Hi,

> On Jan 19, 2021, at 11:35 , Van Snyder via J3 <j3 at mailman.j3-fortran.org> wrote:
> 
> 
> libm isn't the only collection of freely-available reliable mathematical software.
> 

The point was never that libm was freely available.
It was, that with then-existing compilation technology,
many, if not most, Fortran programs were linked with libm
every time, because they used some of it.

So why not call the rest of the routines in libm?
Note that Absoft supported (most) libm routines before f08.

The Bessel stuff isn't in the C standard, it was part
of the Bell Labs C.  But libm, in practice, had them.
(Apparently, calculations involving cylinders were popular
at Bell Labs.)

I now believe that the implementation effort, for at least some vendors,
exceeded the benefits to the language as a whole.

Delineation of use-cases might have caught that.

<snip>

> The reason to add Bessel functions was because they appear frequently in mathematical problems, not in operating systems, games or web pages. It has nothing to do with Keeping Up With the Joneses.

Keeping up with C/C++ in terms of math library, for a mathematically-inclined
language such as Fortran, is not just Keeping Up With the Joneses.  It is
keeping up with the market.


--

Cheers!
Dan Nagle