Better URL scheme/name demangling about hyde

[fosterbrereton]

I agree that what we have right now is not pretty. The problem is that those characters are not unreserved according to the URI specification, so we have to mangle them. If you have an alternate solution that is more legible and equally functional, let me know.

[mrzv]

Well, one hack would be to borrow angle brackets from unicode. U+3008 and U+3009,〈〉. From my cursory reading, they are allowed in URIs, but not in C++ identifiers. (This could be wrong, I didn't dig too deep.) There are some obvious downsides, but it's a solution.

[mrzv]

In the same spirit, but looking a little bit better are small less and greater unicode signs, U+FE64 and U+FE65, ﹤﹥.

[fosterbrereton]

Originally I was going to restrict the characters to the URI RFC, which requires that any non-unreserved characters be encoded. This would also require Unicode characters to be UTF-8 encoded and then escaped. However I am starting to rethink my decision on this, and am willing to consider using the IRI standard for naming instead. That would give us a lot more flexibility in terms of the characters we can use in the paths, and would let us scale back the kind of mangling we'd need to do. I appreciate your input; let me think about it a bit more.