On 3/21/13 7:01 AM, Anne van Kesteren wrote:
On Wed, Mar 20, 2013 at 10:46 PM, Boris Zbarsky <bzbar...@mit.edu> wrote:
>
My ongoing assumption has been that we want to serialize to punycode always in the API
Hmm. That _does_ suck if you want to show location stuff to the user, though.... (though a proposed ACE-to-Unicode API would somewhat help, if you remember to use it).
as the Unicode form is very much a UI decision and we do not want to leak that detail to web developers.
You mean the decision as to when exactly to use the Unicode and when to use ACE? In what way is it a UI decision (apart from browser URL bars having anti-phishing logic that falls back to ACE when they detect "weird" chars in the URI)?
(We do at some point want to expose an API to convert one to the other, irrespective of UI considerations.)
Agreed, but that doesn't help answer the question of what authors should get in the DOM.
Or put another way, it seems like we should be trying to use IRIs until we can't for some reason, and I see no a priori reason that location.href can't be an IRI...
-Boris
