23958 2013-12-02 20:19:38 +0000 Percent-decode on non-ASCII input 2014-01-14 16:11:34 +0000 1 1 1 Unclassified WHATWG URL unspecified PC Linux RESOLVED FIXED P2 normal Unsorted 1 simon.sapin annevk mike sideshowbarker+urlspec oldest_to_newest 96982 0 simon.sapin 2013-12-02 20:19:38 +0000 http://url.spec.whatwg.org/#host-parsing [[ Let host be the result of running utf-8's decoder on the percent decoding of input. ]] Percent decoding is defined for "a string using code points in the range U+0000 to U+007F", but there is not guarantee that input does not contain other code points at that point of host parsing. What happens to non-ASCII code points? 97034 1 annevk 2013-12-03 13:14:40 +0000 How do you get non-ASCII there? 97036 2 simon.sapin 2013-12-03 13:29:18 +0000 new URL('http://☃/') 97037 3 annevk 2013-12-03 13:33:32 +0000 So one option would be to percent escape during the host name state. But that is not the only entry point to the host parser. Hmm. 97144 4 simon.sapin 2013-12-04 19:01:36 +0000 The host parser should UTF8-percent-encode, just before it percent-decodes in step 3. 98429 5 annevk 2014-01-14 16:11:34 +0000 https://github.com/whatwg/url/commit/3cfaa1779bfb9a3ba2b907a5802ca0251ca9a7e6