Re: Fork

[Paul Brook <paul at codesourcery dot com>]

On Monday 24 May 2004 00:10, William M. Quarles wrote:
> Janne Blomqvist wrote:
> >> It would be nice if GFortran included a
> >>commandline option for Essential FOTRAN 90/95 compilation.
> >
> > In principle I like this idea. There is however the problem of
> > choosing which subset to support, F, ELF90 or some gfortran specific?
>
> As far as I understood, there isn't much difference between F and ELF90.
>     But it wouldn't be too much toruble include both.
>
> As for a GFortran specific one, I would imagine that some union of the
> two subsets plus the 95 additions would suffice (although I think F
> includes both WRITE and PRINT, which are redundant).  If there are any
> new features added in FORTRAN 90 that were not included in these
> subsets, then those obviously should be included in a Gfortran subset.

If F and ELF are indeed subsets of F9x then any f95 compiler should already be 
able to compile them, right? Even if they aren't strict subsets I'd be happy 
for someone to add the missing bits as an extension.

However I'm less keen on enforcing the subset rules within gfortran. This has 
the potential to become a maintenance nightmare. We already have at least 
three different proposed subsets: F, ELF, and "The nice bits of 
gfortran+extensions".

I'd prefer to stick to enforcing the official ISO Fortran standards, which 
while not perfect are at least reasonable well defined.

Wouldn't a F/ELF conformance checker be better written as a source 
preprocessor? It could then be bolted onto the front of any f9x compiler. As 
you're only handling the nice bits of fortran writing the parser shouldn't be 
that hard. There are several similar "lint" checkers available for C code.

Paul