Boost logo

Boost :

From: Beman Dawes (beman_at_[hidden])
Date: 2000-12-07 14:14:07


At 09:50 AM 12/7/2000 -0800, Gary Powell wrote:

>> I'm in violent agreement here, it is a *lot* easier to have the names
in
>> boost.
>>
>This does not make it easier for library creators.
>
>Since there is no registry for official names and I've got several large
>libraries, I either suck up all available good names or nest them within
my
>own namespace. Or fight with other libraries to be first and stake out my
>claim.
>
>Tuples is a good example. Python_c++ has them, as does LL.
>
>So rather than try and manage the individual names I'd rather manage
>namespace names and let library builders have total freedom within their
>namespace.
>
>So if we want boost level names I want only a relatively few of them. And
>they should be extremely useful so that no one minds that they are now
>effectively reserved.

It seems to me that there is a useful distinction between larger libraries
and smaller ones. It makes sense for larger libraries with lots of names
to be in a sub-namespace. But for small libraries, particularly those of a
very general nature, having a nested namespace doesn't seem like it buys
much. There are also the compiler issues with nested namespaces.

Of course if you buy those arguments, it would mean libraries like the
boost graph library should be in a nested namespace... We really do need
to decide on a policy, publish it, and try to follow it.

--Beman


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk