On 02/17/2010 10:25 PM, Andrei Alexandrescu wrote:
What evidence do you have that it's widely useful?
I use it once in a while in python, evidence enough? :)
Well I understand you don't agree from the frame you're now in, but you haven't heard my argument yet. Under such conditions you shouldn't say "can't"! Here goes my argument. 1. An array literal has no name and cannot be aliased, hence it's private to the implementation. The implementation is therefore free to induce structure on the searched elements, e.g. by sorting the array or using a hash. 2. The size of an array variable must be conservatively assumed to scale with the size of the input. The size of an array literal scales with the size of the program, or in the worst case (code generation) with statically-known inputs to the program.
I don't see how this is any argument against opIn_r for array variables as well.
I mean, the argument against it seems to be to make it more difficult to do a linear search over an array, which seems rather backwards to me.
