[boost] Re: config.h[pp]

From: Beman Dawes (beman_at_[hidden])
Date: 1999-07-15 19:07:09

At 12:00 PM 7/15/99 -0500, Andy Glew wrote:

>> > + namespaces - hah! it's fixed now!
>> >
>> Actually, not.
>>
>> Well, the fix may be partial. That was the problem I had with
operators.h.
>> ... The problem goes away if you take
>> everything out of the boost namespace.
>
>Sigh.
>
>So this provides a more specific question or instance of the
variation problem:
>
>Should Boost avoid using namespaces, because they (still) seem to be
>broken with one of the most popular almost standards conforming
compilers,
>EGCS/G++?

No, IMO.  Namespaces are mainstream.  Most compilers have had them
for at least a couple of years and/or releases now.  There have been
no complaints from any Boost users that I know of, and all the
current Boost libraries use namespaces.  

>Should Boost provide varieties which pollute the global namespace,
as well
>as ones which do not, e.g.:
>
>        boost/no-namespace/operators.h:
>            ...
>
>        boost/operators.h:
>            namespace boost {
>                #include "boost/no-namespace/operators.h"
>            }
>
>Or should we wait for (or hope that one of us gets off our duff and
does it
>for everyone) GCS/G++'s namespace implementation to be fixed?
>
>===
>
>The various brokennesses of namespaces were one of the things that
depressed
>me most about EGCS/G++.  IMHO, namespaces are essential to well
defined
>libraries that can interoperate.

Agreed.

--Beman

------------------------------------------------------------------------

eGroups.com home: http://www.egroups.com/group/boost
http://www.egroups.com - Simplifying group communications