|
Boost : |
From: Brian McNamara (lorgon_at_[hidden])
Date: 2003-11-14 12:01:47
On Fri, Nov 14, 2003 at 06:23:28AM -1000, David Abrahams wrote:
> Sometimes, I think named conformance is too heavyweight. Is that
Indeed; without language mechanisms designed _for_ this, we must slog
through "simulating" them via template metaprogramming.
> just a side-effect of the way the language is defined?
> http://www.boost-consulting.com/writing/qn.html shows a proposal I
> made to, in part, fix that problem.
I'll have a look.
> YES! "INTRUSIVE METHODS! INTRUSIVE!" ;^)
LOL
> Not really, if library::clonable is a type with a prescribed
> meaning. If it's in the contract that by using it you have a certain
> semantic intention, it's fine.
You're right; I didn't look closely enough at Brock's example.
(I glanced and thought he was saying conformance was "if a class has a
member type named 'clonable'", rather than "if a class has a member type
named 'clonable' which is the same type as 'library::clonable', which is
the name associated with the intention".)
> I think you should look at the concept-checking papers in the pre-Kona
> mailing. What you're proposing is not really a response to that.
Can you point me at these documents, please?
-- -Brian McNamara (lorgon_at_[hidden])
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk