Ownership issues

[RokerHRO]

At the moment ownership aspects are completely undocumented, so I looked into the code:

• url->href is a non-owning a shallow copy of the parameter given to url_parse() and not touched by the remaining code at all.
• all other members of struct url_data are each heap-allocated copies, owned by the struct (and therefore free()'d in url_free()).

I'd like to know which property/behavior is as desired, which is just by coincidence and might be changed in a more optimized implementation (which I already have in mind, e.g. to support key-value access to the content of the query string, incl. percent decoding of non-ASCII data etc.) Different ownership would not affect clients/users that use the data read-only, but clients that changes the member would break.

[jwerle]

@RokerHRO I think if you have some good ideas about memory ownership, feel free to express them in a PR! I am highly confident it will be merged in :)

[RokerHRO]

@RokerHRO I think if you have some good ideas about memory ownership, feel free to express them in a PR! I am highly confident it will be merged in :)

I got some ideas and working on it. :-)