(j3.2006) C629

[Malcolm Cohen]

Yes, I think so.  Removing the requirement makes leaving off the type-spec an 
easy mistake to make (likely to result in memory overwrites later, and not 
detected if compiling without checking).

Given
    CHARACTER(10),POINTER :: x
    CHARACTER(:),POINTER :: y

one expects
    ALLOCATE( name )
to have the same effect as
    ALLOCATE( typespec :: name )
where typespec ***is the typespec in the declaration of name***, not some random 
different typespec with the same type but one or more type parameters different.

Thus
    ALLOCATE( x )
acts as a CHARACTER(10) allocation, and
    ALLOCATE( y )
acts as a CHARACTER(:) allocation, which is not allowed since ":" is not a 
value, i.e. you get an error message.  Just because
    CHARACTER z
has a default LEN=1 for declaration purposes does not mean
    CHARACTER(:) y
should mean "LEN=1" for non-declaration purposes, and in fact it doesn't.  It 
means you have to provide something that specifies it.  Not providing anything 
is different from "providing something".  If you really do want the "default if 
you did not specify LEN" then
   ALLOCATE( CHARACTER :: y )
is available, and here you have told the compiler "yes, I really do want the 
default for declarations even though I wasn't declaring anything".

Cheers,

-----Original Message----- 
From: Van Snyder
Date: ?? 26?5?29? 11:06
To: j3
Subject: (j3.2006) C629

C629 says that if an <allocate-object> has a deferred type parameter,
either <type-spec> or <source-expr> shall appear.

Is that really necessary if the type parameter has a default value?


_______________________________________________
J3 mailing list
J3 at mailman.j3-fortran.org
http://mailman.j3-fortran.org/mailman/listinfo/j3

________________________________________________________________________
This e-mail has been scanned for all viruses by Star.
________________________________________________________________________

-- 
................................Malcolm Cohen, Nihon NAG, Tokyo.