🚀 go-pugleaf

In article 4kTmzbC00iVD0Bmn4x@andrew.cmu.edu, Mathew G Monroe <mmex+@andrew.cmu.edu> writes:
>
>First: Under the current draft standard are all const object only given
>an internal linkage, unless they have previously been given external
>linkage?  I ask because I understood this was the case. 

Yes.

> But, GNU G++
>(at least 2.7.0) doesn't strictly follow this.  For example,
>
>const char *messages[]={"Fun", "Stuff};
>const char fun[]="message";
>
>messages is made external, and fun is internal.  This seems wrong to me,
>both should be internal.

No, GNU is correct. "messages" is not const, so it is made external.
That is "messages" is a non-const pointer to a const char. On the
other hand, "fun" is a const array of chars.  The "const" applies to
the declarator which follows. In the first case, the declator is "*"
and in the second case it is "fun". Pointers and arrays are not
the same thing.


>
>Second: By the current standard is the complex class a template?  I ask
>because old versions of GNU g++ and libg++ has complex not as a
>template, yet g++ 2.7.0 and libg++ 2.7.0 make it a template.

Yes, the current draft makes complex a template. The exact spelling
of the names and whether the spelling "complex" should be a typedef
to something like basic_complex<double> instead of the name of a 
template was still under discussion the last time I checked.

---
Steve Clamage, stephen.clamage@eng.sun.com




---
[ comp.std.c++ is moderated.  Submission address: std-c++@ncar.ucar.edu.
  Contact address: std-c++-request@ncar.ucar.edu.  The moderation policy
  is summarized in http://dogbert.lbl.gov/~matt/std-c++/policy.html. ]

Steve Clamage (clamage@Eng.Sun.COM) wrote:
 
: >const char *messages[]={"Fun", "Stuff};
: >const char fun[]="message";
: >
: >messages is made external, and fun is internal.  This seems wrong to me,
: >both should be internal.
:
: No, GNU is correct. "messages" is not const, so it is made external.
: That is "messages" is a non-const pointer to a const char. On the
: other hand, "fun" is a const array of chars.
 
I'm sure it's intended that "fun" should have internal linkage.  But
from a casual perusal of the draft (April working paper), it seems to
me that "fun" is not a const array of char, but an array of const
char, and that "fun" itself is _not_ const (although "fun[0]" is), and
so it has external linkage.
 
Have I missed something, or is this a mistake in the draft?

-- 
John Rickard
---
[ comp.std.c++ is moderated.  Submission address: std-c++@ncar.ucar.edu.
  Contact address: std-c++-request@ncar.ucar.edu.  The moderation policy
  is summarized in http://dogbert.lbl.gov/~matt/std-c++/policy.html. ]

In article <199510291313.NAA10621@anchovy.atml.co.uk> John Rickard <jrr@atml.co
.uk> writes:
>: >const char *messages[]={"Fun", "Stuff};
>: >const char fun[]="message";

>I'm sure it's intended that "fun" should have internal linkage.  But
>from a casual perusal of the draft (April working paper), it seems to
>me that "fun" is not a const array of char, but an array of const
>char, and that "fun" itself is _not_ const (although "fun[0]" is), and
>so it has external linkage.
> 
>Have I missed something, or is this a mistake in the draft?

You're right that the type of "fun" is "array of const char", but the
definition of "const-qualified" in C++ (unlike in C), when applied to an
array, tests the qualification of the underlying element type.
So the type of "fun" is const-qualified.  It must be admitted that
the text relating to this in the WP [7.1.5.1 and 3.9.3] could be clearer,
but the precise question you bring up was brought before the full committee
and voted on, with the outcome as I described it.

Steve Adamczyk
---
[ comp.std.c++ is moderated.  Submission address: std-c++@ncar.ucar.edu.
  Contact address: std-c++-request@ncar.ucar.edu.  The moderation policy
  is summarized in http://dogbert.lbl.gov/~matt/std-c++/policy.html. ]