On 23/08/2013 16:24, roger riggs wrote:
Hi,
I reviewed the algorithm of the test case and it seems to me that it
produces the smallest number of total items to sort for a given stack
depth.
I ran it with other stack depths and confirmed that the implementation
did
not exceed the new limit.
I was unable to discover a specific explanation or computation for the
current choices to thresholds and corresponding stack depths.
An alternative implementation could use the conservative estimate
and resize the array as needed. It would avoid overallocation of a
temporary in the normal case.
Thanks, Roger
I searched around but don't see any crumbs that explain the current
stack depths.
Anyway, I think what you have is fine and should be pushed.
-Alan
