However, the JS gettters are all over the place... Given this: a = document.createElement('a') a.href = "/ö/o?ö/o#ö/o" console.log(a.hash, a.pathn

However, the JS gettters are all over the place... Given this: a = document.createElement('a') a.href = "/ö/o?ö/o#ö/o" console.log(a.hash, a.pathname, a.search) we get: Browser .hash .pathname .search Chrome #ö/o /%C3%B6/o ?%C3%B6/o Firefox #%C3%B6/o /%C3%B6/o ?%C3%B6/o Safari #%C3%B6/o /%C3%B6/o ?%C3%B6/o Edge #ö/o /ö/o ?ö/o IE11 #ö/o ö/o (ouch) ?ö/o Likewise for domains, some UAs return them Punycoded (but don't provide a decoder), some in Unicode, and PhantomJS passes it percent encoded... It's a big mess. I discovered this while working on a SPA router that can handle routes in either `pathname`, `search` or `hash` mode... The routes are defined as the keys of a JS object, in Unicode in the source, and we have to juggle back and forth to match what `document.location.*` returns.