Re: [ANN] Lua 5.2.0 (alpha) now available

[Enrico Colombini <erix@...>]

On 13/12/2010 9.29, Dirk Laurie wrote:
> The algorithm for #t in ltable.c is (if I read it right):
>
> Consider k, the size of the array part of t.  If t has no hash
> part and t[k] is not nil, we immediately have #t=k (constant time).
>
> Otherwise either k-1 is an upper bound, or k is a lower bound for #t,
> depending whether t[k] is nil.  In the latter case, an upper bound
> can be found by doubling (logarithmic time). Having an upper and a
> lower bound, binary search (logarithmic time) is used to find #t.

So, accessing #t for a mixed table (array + hash) is slower than 
accessing it for a pure array with the same data?

--
  Enrico