Namespace: explicit parsing errors

[ppkarwasz]

This PR contains the part of #452 that is specific to namespace.

It requires the parser to throw an error if a (percent-encoded) solidus / is encountered in any path segment.

[ppkarwasz]

Does this simplify namespaces? This seems like it only makes namespaces more complicated. Unlike subpaths, namespaces do not care about . or .. segments (and to do so would be a breaking change) so there is no reason to treat %2F and / differently.

Originally posted by @matt-phylum in #452 (comment)

[johnmhoran]

@ppkarwasz I'm not sure if you overlooked replacing "solidus" with "slash" or instead are insistent on using "solidus". I have no linguistic objection to that beautiful Latin-based word, but we don't use it anywhere else in the spec -- it's slashes all the way down -- and RFC 3986 similarly uses "slash" not "solidus". Some famous person somewhere once said something like "a foolish consistency is the hobgoblin of small minds", but imho consistency in the terms we use is a distinct advantage.

What do you think?

[ppkarwasz]

I'm not sure if you overlooked replacing "solidus" with "slash" or instead are insistent on using "solidus".

I didn't notice it. Fixed in 9bd2568

Regarding solidus: it is the Unicode name of the character, but I don't really care about the naming as long as it is consistent (which as you pointed out was not).

[johnmhoran]

Thanks @ppkarwasz -- LGTM!