Re: [draft-duerst-iri-bis-06] differences from HTML5 algorithm

On Sat, Sep 12, 2009 at 2:17 AM, "Martin J.
Dürst"<duerst@it.aoyama.ac.jp> wrote:
> On 2009/09/11 23:24, Erik van der Poel wrote:
>>
>> On Fri, Sep 11, 2009 at 2:11 AM, Anne van Kesteren<annevk@opera.com>
>>  wrote:
>>>
>>> On Fri, 11 Sep 2009 10:57:40 +0200, Martin J.
>>> Dürst<duerst@it.aoyama.ac.jp>  wrote:
>>>>
>>>> This essentially says that you MAY send data from a form in the document
>>>> encoding, and was followed well by browsers. It seems that some browser
>>>> implementer along the way extended that to query parts in other URIs
>>>> (which
>>>> don't have anything to do with<form>), and got stuck with it.
>>
>> I don't know which browser version first did this. I don't even know
>> which vendor. It may have been Netscape or MSIE (or ...).
>>
>> But look at it from the point of view of the server. The server cannot
>> tell whether the incoming URI is from an HTML<form>  or an HTML<a>.
>> So it would be nice if the browser treated both the same way.
>
> Of course. But it's not just forms (where we have accept-charset) and and
> <a>. It's also stuff typed directly into an address bar, where there is not
> much choice except to assume UTF-8.
>
> So whichever way, the server may have a tough job.

Yes, that is another source. My guess is that servers will receive
these URLs from HTML <form> and <a> much more often than the UI
address bar, since users don't like to type such long strings, and the
query part is at the end of a pretty long string.

Anyway, we seem to agree that it is too late to change this now.

Erik

>> (Unfortunately, this means that<a href="...">  with query parts that
>> have nothing to do with HTML forms (i.e. name=value[&] pairs) also get
>> converted back to the original encoding, but I agree with Anne that it
>> is too late now to change this.)
>>
>>>> This is especially unlucky as for<form>, you can just say
>>>> accept-charset='utf-8', and get all the data sent in UTF-8,
>>
>> Does this work in all major browsers? (I don't know; I haven't tested it.)
>>
>> Erik
>>
>>
>
> --
> #-# Martin J. Dürst, Professor, Aoyama Gakuin University
> #-# http://www.sw.it.aoyama.ac.jp   mailto:duerst@it.aoyama.ac.jp
>

Received on Monday, 14 September 2009 19:53:18 UTC