CGAL users discussion list

[Sylvain Pion <>]

Ben Supnik a écrit :
> Sorry, what I meant to write was: "I don't think any of the kernels have 
> SLOWER than constanttime operation".

I guess that it really depends how you count.  In practice and in general,
I would not state anything like that.

> Sylvain: in the case of some of the lazy kernels is it possible that 
> time performance of the whole algorithm could be improved by the 
> reduction of entire sets of geometric or numeric operations that can be 
> eliminated when the DAG is evaluated?

It's hard to detect this at the kernel level.  The kernel is meant to deal
with the numeric issues.

> Of course...if this was possible, the algorithm running on top of the 
> kernel would be rather poorly written.

I agree.  Unfortunately, sometimes reality is suboptimal...

> Still, I am excited for the wonderful future of sub-constant time 
> operations. :-)

Wait for quantum computers :)

--
Sylvain Pion
INRIA Sophia-Antipolis
Geometrica Project-Team
CGAL, http://cgal.org/