EXTERNAL: Re: Bug report gcc 4.8

[Mark Rashid]

On Wed, 2014-05-14 at 14:49 -0400, Katherine Holcomb wrote:
> I use lots of advanced features up to and including classes.  I have had 
> no problems with gfortran 4.8, but I avoid pointers (in all languages) 
> as much as possible.


Apologies for misusing the list here. . . but just couldn't resist
adding a further perspective.  I lead the development of a large (0.75M
lines and much more to go) simulation code.  The code is entirely
Fortran 2003, except for a fairly small C++ part that makes OpenGL
calls.  This project started 3 years ago, and was from scratch - no
legacy F77 code involved.  We made the decision from the start to use
gfortran as our standard compiler.  We use every corner of the language,
including all of the major "modern" features, and very extensively.  We
monitor this list and other places for information about what features
work and which ones don't quite yet.  "Grateful" does not begin to
describe how we feel about the gfortran effort.  

I have come to the conclusion that "Fortran-phobia" is more of a
cultural phenomenon than anything else.  The 2003 standard is a hugely
powerful language that is ideally suited to scientific programming in so
many ways, including - and especially - efficiency in human-effort
terms.

Mark Rashid

> 
> On 05/14/2014 02:42 PM, Brown, Gregory C wrote:
> > My company supplies gfortran also and authorizes it for use on every computer around.  But the 'volunteer' status means it's an unsupported product: use at your own risk.
> >
> > Apparently, the catastrophic bug from f90-standard days still present in gfortran is not so objectionable to many people.  I'm pretty sure I'm the only person in my company attempting to use advanced fortran features.  Everyone else is just compiling legacy f77 code. I suspect this situation is widespread: not too many people are using gfortran and exploiting the advanced language features.
> >
> > ________________________________________
> > From: Toon Moene [toon@moene.org]
> > Sent: Wednesday, May 14, 2014 12:31 PM
> > To: Brown, Gregory C; Tobias Burnus; W Spector; Dominique Dhumieres; fortran@gcc.gnu.org
> > Cc: sgk@troutmask.apl.washington.edu
> > Subject: Re: EXTERNAL: Re: Bug report gcc 4.8
> >
> > On 05/13/2014 09:53 PM, Brown, Gregory C wrote:
> >
> >> 4) I realize that gfortran is open source and done by volunteers (and I realize I ought to contribute in some way if I use it) -- but I must ask -- would gcc C++ have the same problem? Must C++ programmers rely on commercial compilers to get all the language features they need to work correctly? This is the argument that my company gives me for NOT wanting to use open source at all. Not well supported.  Or is it just because (as I receive the charge) nobody in their right mind but me wants to program in modern fortran?  A C++ programmer friend at work tells me that in some sense the gnu C++ compiler is more 'pure' standardwise.  Is that not the case with gfortran?  Or again is this just the network effect: too few people using fortran.
> > The "volunteer" status of gfortran doesn't stop Cray, Inc. (cray.com)
> > from supplying it with its supercomputers as one of the supported
> > Fortran compilers.  I have access (as part of my day-time job) to one of
> > these machines (at the European Center for Medium Range Weather
> > Forecasts - ecmwf.int).
> >
> > Kind regards,
> >
> > --
> > Toon Moene - e-mail: toon@moene.org - phone: +31 346 214290
> > Saturnushof 14, 3738 XG  Maartensdijk, The Netherlands
> > At home: http://moene.org/~toon/; weather: http://moene.org/~hirlam/
> > Progress of GNU Fortran: http://gcc.gnu.org/wiki/GFortran#news
> 
>