On Mon, Sep 27, 2010 at 12:54 PM, David Abrahams <dave@boostpro.com> wrote:
At Mon, 27 Sep 2010 11:40:31 +0800, Dean Michael Berris wrote:
The way concepts are done now is to define the syntactic requirements (not so much the semantics)
I think I know what you mean, but I wouldn't say it that way.
I was looking for a better way to say it, but the way it is done now with Boost.Concept_check is pretty much by trying out the interface of the type being checked whether it models a given concept. I'm not sure how to define a model using pure C++ code to say that a given expression should have O(n) computational complexity with regards to `n` being the size of the input -- of course, you can define this in documentation, which I'm not sure really helps with better compiler error messages. :D
Notwithstanding the fact that the compiler can never verify semantic conformance to a concept, the semantics are a required part of any concept definition. If your concept omits semantics, it's broken.
Definitely true in the strict sense that Concepts in the STL documentation are defined. I was referring to the way it's done with Boost.Concept_check and with regards to better compiler error messages. :) -- Dean Michael Berris deanberris.com