Aleksandar Donev wrote:
> Drew McCormack wrote:
>
>> I realize they are coming in the latest standard, but it will be
>> several years before there will be widespread support in compilers,
>> I expect.
>
> That depends on user demand. Polymorphic variables (pointers and
> dummies) are not *that* hard to implement. Especially polymorphic
> pointers, and especially CLASS(*) ones. It is harder to implement
> type-bound procedures (dynamic dispatch). Then harder to implement
> parameterized derived types. Etc. (although how hard may depend on
> previous compiler infrastructure of course).
I would caution against assuming this is the order in which compilers
will implement 2003 features. Much of the order is customer driven and
there are other features, such as C interoperability, that were a lot
higher on the list we have. (I'd also question Aleks' difficulty
ranking, but that's a separate issue.)
>
>
> So it does not have to take "several years". It is possible in fact to
> implement the features piecewise, and in fact I hope this is what will
> happen. I would actually hope something that is a technical report,
> and not formally part of F2003, gets implemented first: submodules!
> And maybe even some of the simpler features we are adding now to F2008.
To the extent that some of the "features" in 2008 are already widely
implemented as extensions, there may be no wait at all. As for the
others, customers will tell vendors what the priorities should be (in
their opinion, at least) and some 2008 features may rate higher than
some of the 2003 features.
Cheers,
Bill
>
>
> Aleks
--
Bill Long [log in to unmask]
Fortran Technical Support & voice: 651-605-9024
Bioinformatics Software Development fax: 651-605-9142
Cray Inc., 1340 Mendota Heights Rd., Mendota Heights, MN, 55120
|