W3C home > Mailing lists > Public > ietf-http-wg@w3.org > October to December 2010

Re: Issue 261: Check for requirements backing test cases, was: Comments on draft-ietf-httpbis-content-disp

From: Adam Barth <ietf@adambarth.com>
Date: Tue, 2 Nov 2010 09:46:15 -0700
Message-ID: <AANLkTik8JK6653CGj-VqpMO2pV_kvG_DmQ-rZOgDfzLE@mail.gmail.com>
To: Julian Reschke <julian.reschke@gmx.de>
Cc: Mark Nottingham <mnot@mnot.net>, httpbis <ietf-http-wg@w3.org>
On Tue, Nov 2, 2010 at 4:11 AM, Julian Reschke <julian.reschke@gmx.de> wrote:
> On 02.11.2010 09:56, Adam Barth wrote:
>> ...
>> I'm more interested in the invalid header field instances.  This
>> document doesn't explain how to parse them, much less how to process
>> them.
>> ...
>
> No, it doesn't (and that's a separate discussion).
>
> While we are at it, let me explain why <http://greenbytes.de/tech/tc2231/>
> has tests for invalid header fields in the first place. After all, if I
> don't really want to specify how they are processed, right?
>
> The reason why there are there is that they serve the purpose of observing
> whether there is any kind of interop in UAs. If there was, it would be
> interesting to see whether that is by accident, or because existing content
> actually requires it.

While I appreciate the implied complement that browser vendors are
good at fixing interoperability and compatibility problems, I think
you're giving us too much credit.  For example, my understanding is
that there's a bunch of content in Asia that requires the nutty
%-decoding user-default-charset handling.  Because a number of
browsers have effectively chosen not to compete in this marketplace,
they lack this behavior.  Therefore lack of unanimity among browsers
doesn't necessarily imply that there isn't existing content that
requires the behavior.

Put another way, if all the browsers had an unlimited amount of data
at their disposal, they might make different implementation decisions.
 For example, once we gathered a metric ton of compatibility data
about what sorts of content sniffing were actually required to
compatibility render the web, we were able to align browser behavior.
Some of the common idiosyncrasies turned out to be useful, even though
they weren't unanimous.

> I imagine that it is controversial how I rate the results. The idea is to
> say "pass" when the header field is ignored, "warn" when something happens
> that could be considered harmless, or "fail" when something serious happens
> (like UA crashing).

A UA crashing certainly sounds like failure.  :)

Adam
Received on Tuesday, 2 November 2010 16:47:26 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:51:32 GMT