Re: [http-state] non-ASCII cookie values (was Re: Closing Ticket 3: Public Suffixes)

[Adam Barth <[e-mail]>]

On Tue, Feb 2, 2010 at 11:34 PM, Julian Reschke <[e-mail]> wrote:
> Adam Barth wrote:
>> ...
>> I've added some basic tests using non-ASCII characters (in this case
>> Chinese): charset0001, charset0002, charset0003.  IE, Firefox, Chrome,
>> and Opera handle these cookies fine.  Safari does something strange
>> that I haven't tracked down yet.
>> ...
>
> Could you please elaborate on what means "handles fine"? After all, it's not
> specified, right?

I mean the bytes are treated opaquely, which is what the draft
specifies.  The draft specifies what to do with every sequence of
bytes a user agent can receive from a server.

On Tue, Feb 2, 2010 at 11:38 PM, Maciej Stachowiak <[e-mail]> wrote:
> I assume he means that the same bytes that come in as part of the value in a Set-Cookie header get sent out in a future Cookie header, even if they are non-ASCII.

Precisely.

> I am curious if the Cookie name, and not just the value, can contain non-ASCII in existing implementations.

Yes, it can.  charset0002 tests a non-ASCII name and value.
charset0003 tests a non-ASCII name, value, and custom attribute, all
of which acts naturally in IE/Firefox/Chrome/Opera:

== charset0001 ==
Set-Cookie: foo=春节回家路・春运完全手册
Cookie: foo=春节回家路・春运完全手册

== charset0002 ==
Set-Cookie: 春节回=家路・春运完全手册
Cookie: 春节回=家路・春运完全手册

== charset0003 ==
Set-Cookie: 春节回=家路・春运; 完全手册
Cookie: 春节回=家路・春运

Adam