I'd be unhappy about repeating subroutine arguments. In my (rather old)
paper http://www.nag.co.uk/other/ff96/pap_bray.html which focussed on
compilation issues I rejected manually coded interface blocks because of
the problems of double maintenance.
I can see how compilers might analyse a submodule system to calculate a
more limited set of dependancies, but if all the information is in one .f90
file, how do you avoid any work on it producing a new .o and .mod file, and
so trigging the mindless make.
If submodule information is only useful at link stage, won't you delay the
argument checking benefits of putting subroutines in modules until the
link? This will slow down code development compared with the current system
where you can add an argument and keep recompiling and fixing warnings
about its use as they occur.
I don't feel happy enough with the proposal to do a paragraph by paragraph
commentary I'm afraid
John
--
John Bray, Numerical Weather Prediction Tel: +44 (0) 1344 854035
Room 337 [log in to unmask]
The Met. Office http://www.met-office.gov.uk
London Road, Bracknell, RG12 2SZ, UK http://www.jrbray.org.uk
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|