W3C home > Mailing lists > Public > ietf-http-wg@w3.org > October to December 2011

Re: lower casing host names

From: Roy T. Fielding <fielding@gbiv.com>
Date: Sat, 31 Dec 2011 11:38:24 -0800
Cc: Anne van Kesteren <annevk@opera.com>, ietf-http-wg@w3.org
Message-Id: <8152EA66-C596-435D-AABD-BD59594284B0@gbiv.com>
To: Adam Barth <w3c@adambarth.com>
On Dec 31, 2011, at 10:19 AM, Adam Barth wrote:

> On Sat, Dec 31, 2011 at 1:45 AM, Anne van Kesteren <annevk@opera.com> wrote:
>> On Sat, 31 Dec 2011 10:27:44 +0100, Daniel Stenberg <daniel@haxx.se> wrote:
>>> On Fri, 30 Dec 2011, Adam Barth wrote:
>>>> I suspect it's not really something for HTTPbis to concern itself with,
>>>> however.
>>> 
>>> I am concerned because when the three most used browsers do this it means
>>> lots of people will test their sites and servers with them and everything
>>> will be fine and dandy even when they lazily don't do the right thing - and
>>> this will force other HTTP implementors to go this way too sooner or later.
>>> Pretty much exactly how we ended up with how cookies work.
>>> 
>>> These three browsers then apparently do something with HTTP that isn't
>>> mentioned (or referred to) in httpbis. Such hidden knowledge isn't good for
>>> a protocol spec imho.
>> 
>> You can (and I think we have) run into the same kind of issues with order of
>> HTTP headers, HTTP header casing, etc. HTTP isn't very conservative in what
>> clients have to do, while it probably should be given that there are only so
>> many clients versus many many servers. Although not protocols, we learned
>> the same lessons with e.g. CSS, the DOM, and HTML. (Also known as Postel's
>> law.)
> 
> I suspect we'll eventually want to write a document that contains all
> this "hidden knowledge."  I suspect HTTPbis will call that document a
> "browser profile" of HTTP, which seems fine.  Not all of these
> constraints apply in every situation in which HTTP is used.

Now that we know this isn't just a one-browser issue, we should come up
with a suitable description of it for the Host header field.  Note that
this does break some expectations by server owners (namely, that their
trademarks won't remain in the proper capitalization when used in a URI
on the Web).  I am fine with that, provided what we document is an
accurate description of what some user agents do (and preferably why).

....Roy
Received on Saturday, 31 December 2011 19:38:50 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:51:51 GMT