[Not a review, but a response to one of Paul's comments]
On 23 Apr 2014 at 14:33, Paul A. Bristow wrote:
Though I'm not enthusiastic about the name of namespace boost::typeind:: but I don't have any much better ideas. Perhaps someone else has?
To clarify the issue to everyone else, TypeIndex is in a rather unusual bind with regard to what namespace to use. Ideally I think we'd all agree that boost::type_index would be best, but then one of the issues raised last peer review was that there shouldn't be a boost::type_index type as could conflict with std::type_index, and if we hold that to be wise, then surely the same rationale applies for a boost::type_index namespace as well.
So we end up with boost::typeind, which I don't think makes anyone happy, but it is safe.
I don't suppose boost::typeindex would be an improvement? That name at least immediately suggests to the reader which library to look up. Whichever name is selected, both the documentation and code comments at the namespace definition should explain, as above, why the author intentionally avoided type_index.