Re: RFC resolving typespecs of the kind "typespec function ...."

["Paul Richard Thomas" <paul dot richard dot thomas at gmail dot com>]

Tobias,

> >
> > Any better ideas, comments, suggestions?
> >

I got to thinking about this last night and just now put my thoughts
into the attached patch.

Note that this patch is completely untested; in trying to compile it,
I ran into the out-of-date mpfr syndrom.  I have left gfortran
building but now have to get on with something else.

The (intended!) logic is that deferred KINDs get tried just once after
the last ST_USE and that deferred TYPEs get tried once at the end of
the specification statements.

The ring-fencing of the state verification is probably unnecessary but
I thought not to confuse things in the first instance.

I hope that it helps..... I'll investigate the horrors that Dominique
uncovered tonight:)

Cheers

Paul

Attachment: parse.diff
Description: Text document