Re: sequences

On Tue, Jun 15, 2004 at 03:20:57PM +0100, Malcolm Wallace wrote:

Ross Paterson writes:

...

Another way to use it is to prune out the dominated implementations, and then advise people to choose the implementation that accelerates the operations they care about and as few others as possible (because they'll pay for the extras in higher constant factors).

I would advise against pruning out the 'poor' implementations. Negative results are valuable. Keeping them documented will discourage people from reinventing the (rusty, square) wheel!

OK, then keep them in an attic, but not in the public interface.