"Jonathan Wakely" <cow@compsoc.man.ac.uk> wrote:
On Mon, Oct 11, 2004 at 12:32:42PM -0600, Jonathan Turkanis wrote:
Dinkumware is producing an implementation of TR1 from scratch. In the process, Pete Becker has apparently discovered some small problems with the specifications and some improved implementation techniques, so I don't think independent implementations are such a bad idea.
True. There would be nothing to stop libstdc++ doing an independent implementation later, but for now I think the idea is to put together as much as possible, to have _something_ there to experiment with.
I definitely didn't mean to imply that *all* implementations should be from the ground up. The Boost implementations are pretty good ;-) At some point, though, I'd like to see at least one other independent implementation, in addition to Dinkumware -- preferably more. Jonathan