Re: Rice distribution

[Paolo Carlini <paolo dot carlini at oracle dot com>]

Hi,

On 09/08/2012 01:46 AM, Gabriel Dos Reis wrote:
> On Fri, Sep 7, 2012 at 6:21 PM, Ulrich Drepper <drepper@gmail.com> wrote:
> > On Thu, Sep 6, 2012 at 6:36 PM, Paolo Carlini <paolo.carlini@oracle.com> wrote:
> > > I get your point, but to be honest I'm not 100% sure the issues in this area
> > > are so simple. In my opinion, conservatively, before adding further
> > > non-uglified names to that namespace we should find the time to re-analyze
> > > in detail:
> > >
> > >      http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24537
> > I haven't seen much of a discussion there.  Let me ask, what is the
> > real problem?  I doubt macros should be considered a problem.  As said
> > before, the chosen, new names are long and really shouldn't conflict.
> > In the C library additions are made without leading underscores and
> > there really haven't been any big problems.  And macros are much more
> > frequent in C.
> >
> > The only possible problem I can see is if we add names without
> > underscores to __gnu_cxx and they later are used, perhaps in a
> > slightly changed form, in std.  What if programs use
> >
> > using namespace std;
> > using namespace __gnu_cxx;
> > beta_distribution<> v;
> >
> >
> > Is this the problem people are afraid of?
> I am inclined to tell people "don't do that if it hurts.
> Bring into scope what you really need; not a
> blanket statement."
Looking forward, when we break the ABI, I think we really want to make 
sure that the implementation details aren't in the same namespace of the 
extensions. I agree that if we are very careful with that, things should 
be basically fine. What we'll do with the allocators, I don't know.

Paolo.