Re: Comments on draft-ietf-httpbis-content-disp

On Mon, Nov 1, 2010 at 6:58 PM, Bjoern Hoehrmann <> wrote:
> * Adam Barth wrote:
>>My understanding is that the %-encoding behavior is needed to compete
>>in the Asian market, not the German market.
> Yes, and once you can tell us of popular asian web sites where users
> download a lot of files that get garbled file names in Firefox et al.
> but not in Internet Explorer and Chrome, and that adopting their be-
> havior does not cause problems on other popular web sites for them,
> then you can be sure that Firefox and Opera and Safari and many other
> applications will be updated to match the behavior of Internet Ex-
> plorer, and the HTTPbis Working Group will document that behavior in
> their deliverables; it'd be demonstratively needed, after all.

This seems backwards to me.  You're saying that once we solve the
interoperability problem, then we can write a specification to
document that fact?  Is not the purpose of a specification to make it
easier for folks to write interoperable implementations?

> Without that, the triage teams for those applications will find more
> important bugs to fix ad infinitum, meaning the applications will not
> be updated, and any specification contrary to their behavior will de-
> scribe a fiction, which is worse than not mandating a particular hand-
> ling for percent signs (or the many more, far less important, cases
> you want the specification to address aswell). Philosophical debates
> about the abstract needs of absent browser developers are nice and all,
> but getting application developers to change their code requires good
> evidence; that you are not providing.

If no one is interested in interoperability, then I suggest we turn
out attention to other matters where our standardization efforts can
be productive.

> This is an IETF Working Group; rough consensus is clearly not on your
> side for the moment, so show us the running code. Really do, I'd like
> nothing more than an IETF specification that tells you all the things
> that you really need to know. Don't expect us to act on your hearsay.

Here's some running code:


Received on Tuesday, 2 November 2010 03:15:10 UTC