Arkadiy Vertleyb <vertleyb <at> hotmail.com> writes:
"Joaquín Mª López Muñoz" <joaquin <at> tid.es> wrote
Why not use an MPL sequence of indexes in the first place?
[...]
MPL is an extremely powerful library, but I honestly don't expect it to to be part of the common knowledge shared by the C++ community.
I don't quite agree with you here... Meta-programming has become a major part of C++, and allowing to specify a list of types is going to be more and more common thing. It may be expected in the future by more and more people (as it is expected by some people already) that, whenever multiple types are used, they have to be combined in a standard type sequence. MPL sequence is becomming such standard (as STL has become).
Well, our opinions differ. Just out of curiosity, ask your fellow programmers if they know what MPL is. I have done that, and hardly anyone knew about Boost, let alone MPL. Anyway, please read on. [...]
One compromise would be to use the MPL sequence concept, and have indexed_by implementing it (say derive it from mpl::vector). At least then there would be no need in the MPL converter.
I've done some tweakings and your proposal works. Besides, it makes sense. To sum it up, one would still be allowed to write multi_index_container < employee indexed_by< ...
*and* also multi_index_container < employee mpl::vector< ...
or whatever MPL forward sequence. OK, it seems to me a good compromise. If nobody's against it it'll go in the post review version. Thanx for discussing! Joaquín M López Muñoz Telefónica, Investigacón y Desarrollo