needless memory allocations about ponder

[billyquith]

Thanks for the feedback. Yes, that is something I'd like to deal with at some point. You get the same problem when declaring scripting (e.g. Lua) interfaces. Also, all of the string compares are not efficient.

This library is a fork of CAMP and other users attempted to fix this problem by using hashing, e.g. refer to 16tons/camp@5e382f9

I forked from the source version though as I'd like to tackle the problem slightly differently. I was thinking of adding an "identifier trait" which could be overloaded depending on the project. I've previously used cached strings and hashes, which is much more efficient, and I recently found a nice implementation of this on GitHub: https://github.com/foonathan/string_id

I wasn't aware of string_view, thanks I'll keep an eye on that, a useful concept, and perhaps part of a future solution.

I'd also like to take a closer look at the allocations at some point, the number, and space used. I consider the library in "beta" at the moment. It "works" but needs a good testing in reasonably complex real project. Started doing this in Gwork: https://github.com/billyquith/GWork/tree/ponder

[billyquith]

This is implemented in the 1.3 branch. I have used traits for the identifier types so that the types may be changed in the future, e.g. EASTL, or string_id.

Thanks for the feedback.