[prev in list] [next in list] [prev in thread] [next in thread] 

List:       gcc-fortran
Subject:    Re: STRUCTURE and RECORD construct addtion
From:       "N.M. Maclaren" <nmm1 () cam ! ac ! uk>
Date:       2013-12-02 18:01:01
Message-ID: Prayer.1.3.5.1312021801010.1154 () hermes-2 ! csi ! cam ! ac ! uk
[Download RAW message or body]

On Dec 2 2013, Paul Richard Thomas wrote:
>
>Nick beat me to it by a few minutes!
>
>Such a patch would not be welcome by me either.  Convert to standard
>conforming code rather than potentially wrecking the compiler.  After
>all, the gfortran documentation points out that there is one to one
>mapping between the two.

It's a while since I looked, but I think that there are uses that would
be very hard to convert (there certainly are with UNION).  However, it
is precisely those cases that would be hardest to implement, would be
most likely to break something else, and are bad software engineering.
As you say, the most common uses just map one to one.

What I would do is to write a Python tool to do the conversion, which
would handle all of the clean cases (and almost all the uses) and give
an error message for the rest.  Parsing Fortran enough to do that isn't
hard.  Other people might use Haskell or even Perl :-)

If someone had the energy and time to write distributable tools, they
would be useful adjuncts.  Toolpack is no more and even NAGWare is no
longer being distributed!


Regards,
Nick Maclaren.



[prev in list] [next in list] [prev in thread] [next in thread]