Re: comp.lang.fortran thread

["N.M. Maclaren" <nmm1 at cam dot ac dot uk>]

On Dec 17 2014, Damian Rouson wrote:
> You might be interested in the following thread on comp.lang.fortran:
>
> https://groups.google.com/forum/#!topic/comp.lang.fortran/2ZFxACIOv2E
>
> In particular, Wolfgang Kilian posted the following question on December 
> 16:
>
> "More important is a standard location for installed .mod files in the 
> file hierarchy, for the usual OSs, given the fact that those files are 
> compiler- and version-dependent and should be naturally associated with 
> the corresponding library code.  For instance, if there is some 
> agreement on this in the GNU context, maybe the gfortran team could put 
> a recommendation on their website?" 
>
> Please see the above URL for more context and background.

Er, I don't know how to say it and not sound like the arrogant old
fogey I am :-(

There isn't the chance of a flea in a furnace of sorting this one out,
short of some full deity (mere semi-deities will not do) issuing precise
instructions on how to do it and immolating anyone who dares to dissent.

This issue predates Unix and C, has rarely been resolved on any single
general-purpose system, and has never been resolved on more than one.
There isn't any need to make such a spectacular mess of it as almost
all modern Unices do, but sorting it out makes undoing Gordian knots
while simultaneously herding cats look easy.

For solely gfortran under Linux-like Unices (which most definitely do
not include AIX), it MIGHT be possible to improve the matter, but I
wouldn't bet on even that.  If you also wanted to allow for some of the
more advanced use of modules (Python numpy springs to mind as being
perverse enough to be a new low even for me), I wouldn't know where to
start.


Regards,
Nick Maclaren.