Re: Feedback on __mt_alloc?

[Stefan Olsson <stefan at snon dot net>]

Phil Edwards wrote:

> On Thu, Apr 24, 2003 at 08:19:42PM +0200, Stefan Olsson wrote:
>  
>
> > Benjamin Kosnik wrote:
> >
> >    
> >
> > > > Looking at the current stl_alloc.h that would only be __malloc_alloc and 
> > > > __debug_alloc right?
> > > >        
> > > I was thinking of moving them all over, so __pool_alloc would also be fair 
> > > game.
> > >
> > >      
> > Hmmm...
> > ...so stl_alloc.h would only contain __new_alloc as required by the 
> > standard? That would be really nice and clean!
> >    
>
> Let's be more clear about this:  the standard doesn't require "stl_alloc.h"
> nor "__new_alloc".  It just requires that the default container allocator
> use new/delete.
Sorry about the confusion...

>  
>
> > Where would the pool_alloc.h file (along with the other SGI specific 
> > code such as __simple_alloc for _Alloc_traits etc) be included then for 
> > "compability"? In stl_alloc.h?
> >    
>
> I don't think we really gain anything by splitting them up, if we have
> to go to each bits/* header and re-include them.  Or by including all the
> new ones back into stl_alloc.h.
>
> I think... hm.  I think I need to put the __N markup on hold, finish my
> current documentaion updates, and then get back to work on redoing the
> allocators.
Ok. Just let me know if I can help in any way, but the way I interpret 
this is that we should "lay low" until you've done your thing with the 
allocators?

Brgds

/Stefan

--
Experience is what you get when you were expecting something else.