7 Apr
2011
7 Apr
'11
3:19 p.m.
Yes, you are right. I added two benchmark tests to cover both call_out handling using ids and without. The current benchmark produces a case with a lot of collisions, so the results are definitely biased towards the optimization proposed before.
On Thu, 7 Apr 2011, Peter Bortas @ Pike developers forum wrote:
As always when talking performance: I appriciate if new benchmarks are added to the benchmark target. Preferably long before doing any optimizations.