Re: [boost] Re: IndexedSet Review: unified proposal for naming

From: Pavol Droba (droba_at_[hidden])
Date: 2004-04-01 17:23:18

On Tue, Mar 30, 2004 at 09:12:19PM +0000, Joaquin M Lopez Munoz wrote:
> Pavol Droba <droba <at> topmail.sk> writes:
> 
> > 
> > > 
> > > NAMESPACE
> > > * Proposal: boost::container::multi_index
> [...]
> > I have just one remark about the namespace usage. IMHO it is an overkill to
> > provide a special namespace for every container. 
> > I think, that putting this all containers into boost::container namespace 
> > is verbose enough.
> > 
> > There has been very similar discussion about the algorithm namespace 
> > (namely the string algorithm library). Current situation is that everything
> > resides in boost::algorithm namespace and interface names are lifted to boost
> > namespace. This model has been settled as a reasonable compromise.
> > 
> > It is worth to mention, that there are generaty more free stading names in 
> algorithm
> > libraries than in the container ones. So if the name-clashing problem is not 
> here,
> > I don't see it in container case.
> > 
> [...]
> 
> I think indexed_set (or composite_container) cannot
> live without a namespace ot its own. There are many utility
> classes around the container with names like (to pick a few)
> 
> * tag
> * index
> * member
> * identity
> 
> These are *public* classes. Would you choose to have them in
> boost::container?
> 

Ok, I see your point now. Still, I would prefer shallower namespace hierarchy if possible.

Maybe renaming ig these *public* classes whould help. Or just putting these utilities
in an extra namespace, keeping the main container class well accessible.

Well, these are just crazy my ideas, but anyway, the issue is important enough to
be worth of some discussion.

Regards,

Pavol