Hi everyone,

I've been programming in fortran 90/95 for over 5 years now, and I enjoy
very much the new modular and object orientated features that fortran
now includes.

In fact, it seems to me that fortran is at a stage where it should
become easier to start exchanging full modules rather than individual
routines (as in numerical recipes for example). For instance, I've
developed for my purposes a module which defines vectors and all the
operations associated with them. Of course many other people have done
this too. I've also written a couple of modules which have interfaces
defined for f77 libraries (e.g. PGPLOT or CFITSIO).

I have come across various websites where programmers make available the
routines that they use. However, in some cases it is hard to know
whether this code is being maintained, etc., and how popular it is.

As far as I am aware, there does not appear to be any large centralized
repository for fortran 90/95 modules, where users could leave comments
about the modules, rate then by e.g. code quality, speed, etc... I think
this would be invaluable. So some kind of mix between the CPAN archive
for perl, and sourceforge for complete programs or libraries.

On Fri, 30 May 2008 10:05:21 +0100, Thomas Robitaille wrote:

In article 40tude.net>,
Ron Ford nowhere.net> wrote:

Thomas Robitaille spam.com> wrote:

If someone (meaning a group) were to work on this, it would save
a huge amount of duplicated effort worldwide. I hate to think
how many Fortran programmers have written their own procedures
to translate case of strings, or find available unit numbers, or
similar tasks.

That said, I'm not volunteering!

On the technical side, I would guess a rather simple Wiki design (with
mandatory registration) would certainly be quite an efficient way to deal
with the issue. I suppose the main issue would be to have a team
reviewing (and doing minimal testing) on the code posted.

On 2 jun, 16:23, "FX" alussinan.org> wrote:

It certainly is not the technical side of things. The tough
part is to find such a team.

Regards,

Arjen (who has no time to complete this posting ;))

In message
t54g2000hsg.googlegroups.com>,
Arjen Markus writes

I think that quite a few people would be willing to post a number of
their useful routines, if the mechanism for doing it were simple enough.
That would be a good start, and worth trying.

In article , "FX" alussinan.org>
wrote:

In a way, the code peer-testing is not absolutely necessary as a first
step - some people already make their routines available via their
personal web-pages anyway, and no-one other than them has tested those.
Testing could be done indirectly by people who download the routines. If
they find a bug, they can contact the person who maintains that module,
just like standard open-source software. All modules could be required
or encouraged to include a test routine within them, which would also
serve as examples for use.

On 2 jun, 17:47, Thomas Robitaille spam.com> wrote:

FX wrote:

What about just creating a http://code.google.com project
(or sourceforge.net, or whatever) and giving people access
as they decide to commit code?

Regards,