[enriquto]

Not only "climate models"... a large chunk of scipy is just a thin Python layer over decades-old Fortran code. That many physicists chose to use the real thing instead of the fisher-price interface speaks in favor of them.

[analog31]

Actually, as a user of the fisher-price interface, I'm glad that it can bind to C and FORTRAN libraries, so my numerics are based on the highest quality code.

[enriquto]

I'm also a very happy user of the dumbed-down interfaces. But I find it strange when my fellow users are surprised or even horrified at the fact that some people still write and maintain Fortran and C code. Hell the Python interpreter they use is written in C! But apparently if you write in C you are some sort of old man yelling at clouds.

[wrnr]

Gonum wraps the same code, from what i could tell fortran seem to have a neat way to handle more specialised number systems like dual numbers and hyperbolic numbers.

[spacedome]

You can do operator overloading in fortran to add a new number type, don't remember the details but I wrote a quaternion library once.

[cozzyd]

It is sad indeed that so few new languages have a stable ABI like Fortran and C.