On Sun, Jun 05, 2005 at 06:10:00PM +0100, Dave Harris wrote:
In-Reply-To: <20050601173608.GA77601@compsoc.man.ac.uk> cow@compsoc.man.ac.uk (Jonathan Wakely) wrote (abridged):
Actually, it's moot because it's been semi-standardised in TR1, not because you now understand the design goals ;)
Really? My impression was that TR1 had not been reviewed in detail. When
I don't think that's true, see the issues list for a lot of discussion. I'm most familiar with the shared_ptr bits and there were lots of issues raised about the design, many of which Peter Dimov had an instant reply to, since the design and implementation were mature and he had already thought about most of it before work began on TR1. Even so, Christopher was proposing adding user-defined ctors to boost::array, which would be completely incompatible with std::tr1::array and would make aggregate initialisation impossible.
the hashing library was submitted to boost, I understood it to be the same as TR1 yet we found several problems with it. We did not blindly accept it just because it was in TR1.
I didn't mean to suggest that, although I do believe there has been a consensus not to introduce needless incompatibilities between TR1 components and the Boost libraries that inspired them. jon