DTIO, DTP, and Submodules

[Damian Rouson]

One last question before letting this thread go: would it make any sense to suggest to the committee that any of the features named in the subject line be declared obsolescent or to suggest to authors that the features be deprecated along the lines of the “Deprecated features” appendix in Metcalf, Reid, and Cohen? I’m writing another book and considering limiting the coverage to features gfortran supports, but I don’t want to limit myself too much if the feature is likely to be supported by the time the book is published.  At least with regard to DTIO and submodules, I’d hope the features are viable.  DTIO is a nice convenience and submodules can save considerable compilation time on big projects by eliminating compilation cascades.  This was an important concern for one site that recently hosted a short course I taught.  But if I recall correctly, one Fortran standards committee member recently suggested declaring DTP obsolete (though I’m not sure the suggestion was serious).

The number one complaint I hear regarding Fortran is the time lag between the release of the standard and compiler support.  If the most widely used open-source compiler doesn’t support major Fortran 2003 features a decade after the standard’s publication, I wonder if it’s time to throw up the white flag and declare the features too burdensome to implement?  I hope not, but it seems like a question worth asking.

________________________________
Damian Rouson, Ph.D., P.E.
Founder & President, Sourcery, Inc.
510-600-2992 (mobile)
http://www.sourceryinstitute.org
http://rouson.youcanbook.me



On Sep 23, 2014, at 9:42 PM, Paul Richard Thomas <paul.richard.thomas@gmail.com> wrote:

> Dear Damian,
> 
> I have contemplated how to implement parameterized derived types and
> inquiry functions on more than one occasion. We have developed most of
> the machinery that is required in the process of implementing classes
> and finalization.
> 
> However, all of my effort on gfortran right now is devoted to the
> reform of array descriptors for which PR56818 is the meta-bug. See
> https://gcc.gnu.org/wiki/ArrayDescriptorUpdate. Tobias is providing a
> lot of help on this work. When completed this will make gfortran, at
> last, fully F95 compliant! (array pointers to components of arrays of
> derived types do not work preperly.).
> 
> With best regards
> 
> Paul
> 
> On 24 September 2014 00:04, Damian Rouson <damian@sourceryinstitute.org> wrote:
>> GFortran developers,
>> 
>> Out of curiosity, is anybody considering doing any work any of the following three PRs before the trunk enters the next stage?
>> 
>> Bug 48298 - [F03] User-Defined Derived-Type IO (DTIO)
>> Bug 40196 - [F03] [F08] Type parameter inquiry (str%len, a%kind) and Complex parts (z%re, z%im)
>> Bug 52846 - [F2008] Support submodules
>> 
>> It seems these are the only three remaining “major” features required for Fortran 2008 compliance. (As a user, I’m just guessing these are the three remaining PRs that would require the most work.  In case it helps at all to have a user perspective, I would have immediate use for DTIO and submodules.  I’ll probably use derived type parameters eventually, but might hold off on using derived type parameters in hopes that someday the language will have a more complete generic programming capability.
>> 
>> ________________________________
>> Damian Rouson, Ph.D., P.E.
>> Founder & President, Sourcery, Inc.
>> 510-600-2992 (mobile)
>> http://www.sourceryinstitute.org
>> http://rouson.youcanbook.me
>> 
>> 
>> 
> 
> 
> 
> -- 
> The knack of flying is learning how to throw yourself at the ground and miss.
>       --Hitchhikers Guide to the Galaxy