Prev: Floating Point Multiplication - 32 and 64 bit differences
Next: Clarify An Old Question: define a user data containing array with dimensions decided in running time
From: Tobias Burnus on 18 May 2010 13:17
Hi all,

the Fortran 2003 standard (and the Fortran 2008 FDIS) permit that a type
name is the same as a generic name, allowing to write a "constructor" to
use a term from other programming languages.

In words of the standard:

"12.3.2 Specification of the procedure interface" [...]
"A generic name may be the same as a derived-type name, in which case
all of the procedures in the interface block shall be functions."

Additionally, one finds the following constraint:

R457 structure-constructor
is derived-type-spec ( [ component-spec-list ] )

C489 (R457) If derived-type-spec is a type name that is the same as a
generic name, the component-spec-list shall not be a valid
actual-arg-spec-list for a function reference that is resolvable as a
generic reference (12.4.4.1).

I have problems to read the constraint. What happens if the list *is*
resolvable as generic reference? Is this then just not a structure
constructor? Namely, the compiler simply checks whether it matches as
generic reference and if not tries whether it is a valid structure
constructor?

Or does this also impose constraints on which interface generic
procedures are allowed?

I think it is the former, as my interpretation is oscillating between
the two interpretations, I though it is better to ask.

Tobias
From: Richard Maine on 18 May 2010 14:19
Tobias Burnus <burnus(a)net-b.de> wrote:

> R457 structure-constructor
> is derived-type-spec ( [ component-spec-list ] )
>
> C489 (R457) If derived-type-spec is a type name that is the same as a
> generic name, the component-spec-list shall not be a valid
> actual-arg-spec-list for a function reference that is resolvable as a
> generic reference (12.4.4.1).
>
> I have problems to read the constraint. What happens if the list *is*
> resolvable as generic reference? Is this then just not a structure
> constructor?

Correct. I recognize it is sort of a confusingly backwards way for the
standard to express it, but I'm afraid that is typical of
standard-speak. Perhaps a short note would have been in order. Without
this constraint, it would be ambiguous whether one had a function
reference or a constructor.

For a user, I think describing it from a different perspective might be
more clear. Perhaps say that you can write something that looks like
this, which is interpreted as a function reference if possible, or
otherwise as a constructor. But that's not the way the standard usually
expresses things.

--
Richard Maine | Good judgment comes from experience;
email: last name at domain . net | experience comes from bad judgment.
domain: summertriangle | -- Mark Twain
 | 
Pages: 1
Prev: Floating Point Multiplication - 32 and 64 bit differences
Next: Clarify An Old Question: define a user data containing array with dimensions decided in running time