On 11 May 2016 at 08:31, Daniel Stenberg <[email protected]> wrote: > On Tue, 10 May 2016, bch wrote: > >> Going "all-in" or "all-out" may indeed be best route. Regarding this, I'm >> really glad you seem to be seriously considering resisting support for this. > > > Here's a fresh blog post of mine regarding the situation we're in: > > https://daniel.haxx.se/blog/2016/05/11/my-url-isnt-your-url/
Ouch. I now read the 'whatwg' specification. I've written and read specifications and requirements for more than thirty years, and that parsing-algorithm-based way of writing a specification is some of the less useful ideas I've seen so far. In short, very bad. One must wonder who came up with *that' idea. Not someone very experienced, I fear. In any case, for what it's worth, I'm more and more convinced that libcurl should stick to the real specifications, which says, in this case, colon-slash-slash. I cannot fathom why it would be necessary to support something like http:///////path, for any purpose. ------------------------------------------------------------------- List admin: https://cool.haxx.se/list/listinfo/curl-library Etiquette: https://curl.haxx.se/mail/etiquette.html
