Re: vector algorithms and __normal_iterator

[Benjamin Kosnik <bkoz@cygnus.com>]

> I expect there are other places where pointer specialisations are no
> longer used, so should the change be made high up in the call chain
> (in vector above) or lower down (in algorithms)?

I say this is a problem with vector, not the algorithms, per se. It's 
assumptions about what constitutes an interator are no longer true.
Thus, vector should be refashioned, not the algoritms. A bunch of other
vector iterator/pointer issues were just fixed in this same manner. 

Is there any way to test for this? It seems to be mostly a speed thing,
which is where I'm assuming you first noticed it.

Hmm. Gaby also has performance criteria for valarray, so maybe the
testsuite should have some way of dealing with this.

-Benjamin