🚀 go-pugleaf

clamage@Eng.Sun.COM (Steve Clamage) writes:

>In article bbl@voyager.Internex.NET, muzok@msn.com (muzo) writes:
>>
>>is global new/delete in any namespace ? Are there any discussions about such 
a
>>thing ?
>
>Namespaces don't apply to the keywords "new" and "delete" that introduce 
>new-expressions and delete-expressions (any more than namespaces apply
>to "if" or "while").
>
>The entire standard C++ library, which includes the various standard versions
>of operator new and operator delete, is in the "std" namespace.

Although it is indeed stated in the April draft and the September
working paper that all library entities are defined in the
`std' namespace [17.3.1.1/2], it is elsewhere stated that the library
defines `operator new' and `operator delete' in the global namespace [3.7.3/2],

and indeed also that `operator new' and `operator delete'
must be either static class members or global functions
[3.7.3.1/2], and so cannot be defined inside a namespace.
(3.3.4/2 makes it clear that "global" means in the global namespace,
not in any other namespace.)

In other words, the draft standard is inconsistent. 

I think the most likely resolution of this inconsistency is that
17.3.1.1/2 will be changed so that it is clear that the library
definitions of `operator new' and `operator delete' are in the global
namespace.  It is possible that 3.7.3.1 could be changed to allow
allocation and deallocation functions to be defined in a namespace, and
this might well be a better solution, but this would require changes to
the name lookup rules for allocation functions, and I think a change of
that magnitude is less likely at this stage.

--
Fergus Henderson             	WWW: http://www.cs.mu.oz.au/~fjh
fjh@cs.mu.oz.au              	PGP: finger fjh@128.250.37.3
---
[ 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. ]