Bill Long

Just speculation, but I suspect the difference between the underscore usage between maxexponent and set_exponent is that max is already the name of a different intrinsic function.

I generally use whatever the number of spaces created by emacs when I hit "tab". Note that the line length limit will probably get really long (10,000 characters) in the...

Some style suggestions that are probably obvious, but seem to be violated in some user code: 1) Do not use language features that have been deleted (PAUSE, Arithmetic IF, ...)....

I find the proposed 2 space indentation and 100 character line limit very reasonable. The foreign preprocessing notation in the example is horrifying. There is a template proposal before WG5...

Well, the main argument against implementing mean(x) this way is that the meat of the computation is sum(x)/real(size(x)) which is AREADY rank and type independent because the involved intrinsics are...

Comparison with Python or Matlab does not seem relevant. In both cases, native code is very slow, and they have to resort to libraries for any reasonable performance. Most compiled...

Interesting example, though the code for Welford’s method does not vectorize and would not perform well on modern processors. I tried a simple example of the direct code and your...

Interesting lists. Particularly the CP2K and UK Met sites that actively advocate to avoid F2003 as being “too new”. This points out an important issue with coding style documents. It...

Note that the Fortran 202X proposal list includes intrinsic functions for converting Fortran characters to C strings and vice versa.

IGNORE_TKR was an invention of the Cray compiler that predated the TYPE(*),DIMENSION(..) declaration in (what is now) standard Fortran. The sources should probably be "modernized". (Although nvfortran 21.3 falls pretty...