@Aaron said:
<hints id="hah_hints"></hints>Even if I'm wrong, and he was really talking about some as-yet-unimplemented variable-length Pascal string implementation, don't you think that you yourself were being more than a little pedantic to start up with the O(log n) business?
Don't try to stop a theoretician. :)
Also his assumption is not perfectly correct. There are two possible szenarios: First is a practical system where we need a data type that has exactly the size of the maximum string length. Then there is no reason, why such a strlen-method could be Member of O(1) - we can read the size at once.
The second szenario would be a theoretical system, which has no memory limits. This system would also have no limits for the size of its data types, so every finite implementation would again be member of O(1).
marvin