What next Forttran_?

[Terence]

Ok!
I and others wanted to get away from the SPAM on comp.lang.fortran.

Mr B was kind enough to ofer to mediate a new Forum.

But it is only going to be sucessful if there are postings and
discussion.

You all know my hang-up on MS F77 v3.31 as a bug-free compiler that
you can code anything in as long as you attack your problem with the
correct elegant approach.
I have possibly written thousands of Fortran-based applications this
way (over a hundred are still the latest solutions to several
commercial requirements.

Few agree with my approach (but then my clients are worldwide, many
still using third or tenth-hand AT's and 386's, so the softeare has to
run on those too).
So let's change the subject:-

Can I suggest a few important subjects?

1) A discussion on which compilers you consider to be economic and
effective for, say, 90% of Fortran programmers? And WHY?

2) A rehash of all the Fortran language elements you consider to be
a) absolutely necessary
b) important otherwise
c) useful, but there is an altrenative.
d) unwise in retrospect.
e) pointless except academically

3) Similar, but directed at the FORMAT statementt elements

4) Similar, but directed at file access modes.

[e p chandler]

On Sep 5, 12:51 am, Terence <tbwri...@cantv.net> wrote:
> Ok!
>  I and others wanted to get away from the SPAM on comp.lang.fortran.

> 1) A discussion on which compilers you consider to be economic and
> effective for, say, 90% of Fortran programmers? And WHY?

OK. I'll bite. I'm a retro-computing guy. I've just spent the past
week writing programs for an Apple 2e running under Apple DOS! Before
I learned much Fortran 95, I used the 2.95 version of G77. It's
Fortran 77 with some extensions. For most of the programs that I knew
how to write, it was sufficient. It has implicit none, do .. end do,
cycle and exit. It's a single .zip archive. Easy to insall, small
footprint, easy to remove. Considering the old code that is running
around in engineering circles, it might be good enough. As for
hardware, there is the DJGPP toolset. It's not pure MS-DOS, but it
uses a DOS extender. So you need at least a 386.

Likewise, as I mentioned in a different thead, there is Open Watcom.
The good features are that it was developed from a mature commercial
product. Of course there still are bugs. It's cross platform. It does
require at least a dos extender to run but it will compile for targets
down to just DOS. It still does OS/2. (I don't use that, but....). My
major problem with it is that it has its own set of extensions. Its
compiler directives - pragmas - are their own world. Development of C
and C++ are continuing but the Fortran side has become an orphan.
Last, the footprint is getting up there in size.

In the middle I have a copy of DVF 5.0. It's obsolete, but runs on my
old Pentium 120 and K6-2. It has bugs and is not full Fortran 95. Some
of the extensions I like are not there. It's unsupported. There's no
copy protection, like CVF has, so I can move it between machines. I
can extract the components I need to use from the CD to create a
minimal working copy. This does most of the WIndows programming that I
need, except that it's for Win98.

Lastly there are the GNU family compilers. g95 development seems to be
on vacation. gfortran gets built on FX's site infrequently for WIn32
MinGW. Both distributions ae large. Using them inolves yet another set
of options. Documentation is OK but sometimes not wonderful. [Mummit
Khan once chided me for not knowing about the -mrtd switch. Without
knowing about, it was *obscure*.) Only the latest gfortran, not yet
available on FX's site, supports the $gcc attributes construct that is
the clean way of chainging the calling convention on Win32 to stdcall
for interfacing with the WinApi and Visual Basic/VBA. Of course Win64
only has CDECL.
Building g95 or gfortran? Ha ha. Look at all the trouble I had just
building SLATEC.

That's enough of a rant for now. I've got some more Apple 2
programming to do.

---- Elliot

[Beliavsky]

On Sep 5, 12:51 am, Terence <tbwri...@cantv.net> wrote:

I have been programming to the Fortran 95 standard with a few Fortran
2003 extensions (such as allocatable arrays for procedure arguments)
for some time. I consider array operations, free source form, user-
defined types, and optional arguments essential, so I could never go
back to F77. G95 is currently my main compiler, but I occasionally use
gfortran and have used Intel and Lahey in the past. The benefit of
adhering to a standard is that one is not locked in.