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

Re: Content-Disposition next steps

From: Mark Nottingham <mnot@mnot.net>
Date: Tue, 9 Nov 2010 12:53:32 +1100
Cc: HTTP Working Group <ietf-http-wg@w3.org>, Adam Barth <ietf@adambarth.com>
Message-Id: <0E290A31-6275-47A0-A0B8-BC106B07E9DF@mnot.net>
To: Julian Reschke <julian.reschke@gmx.de>
On 08/11/2010, at 12:39 AM, Julian Reschke wrote:

> Hi there,
> the ongoing discussions are very interesting, as they are relevant to most stuff we do in Part 1 through 7 as well.


> My take-away is:
> - there's disagreement about whether we should require specific handling of invalid messages

Actually, I think we have agreement that it should *not* be required; the current discussion is whether it's useful to specify optional handling, and where that should be done. Does anyone disagree (i.e., think it should be required as part of HTTP)?

> - if implementers want to do so, they are free to do that in a separate informational document

Of course.

> But...
> - we should keep in mind that every cycle we spend on this discussion keeps us from doing other stuff, and many over here feel that other things have a higher priority

Agreed. A considerable amount of time and energy has been spent on this, and the opportunity cost isn't zero. I'd much rather see people producing text, giving feedback, and running tests, rather than going around in circles.

> - for C-D, the *real* problem isn't lacking interop for invalid messages, but lack of interop for *valid* messages

There doesn't have to be just one problem.

> So I'd encourage to de-couple this discussion from the actual Content-Disposition spec, and let those who want work on that as a separate activity (the question of whether that should become a WG work item will be interesting).
> I'd like to get C-D to IETF LC as soon as possible, thus get everything *else* we can resolve done in the next days.

Given that we currently have at least two vendors (probably more, I'm just going by public information) working on implementing the draft now (at least partly triggered by us going to WGLC), and it's possible we'll get feedback from them, I'm inclined to pause for a *small* number of weeks, to see if we can learn anything else. 

During that time, Adam can come up with a proposal for error handling, which we'll consider at the end of that period. 

Adam, you can do that either openly on the mailing list, or as a design team (i.e., work with other interested parties separately and bring a proposal forward when you think you're ready). I'd encourage you to do it openly on the list, to get constructive feedback from others as you go (thereby increasing your chances of delivering something people will accept). However, if Adam does this, I'd ask people not to question *why* it's being done. 

If it's done in time, and if we gather consensus on it, it can go into C-D as an optional appendix. If it's not ready in time, it can be published (by the WG or separately) on its own. Likewise, if the WG doesn't gain consensus on it, it can be published individually (in the IETF or elsewhere). 

Right now, I'm thinking three weeks -- November 30th (coincidentally, my birthday). This will give us time to re-publish, gain consensus (or not) and go to IETF Last Call by the end of the year.

Adam, is that workable for you? 

> The currently open issues are at <http://trac.tools.ietf.org/wg/httpbis/trac/query?component=content-disp>, and I believe this list contains open tickets that can be closed as duplicates (Mark?).

I'll go through those on-list separately.

> Also, I'm going to say that I consider the work on tests, and documenting the current UA problems in a single place was a success. We got minor problems fixed in Opera and Konqueror, Mozilla is likely to improve soon, and the Chrome nightly builds now have RFC 5987 support.

Indeed, and I'd like to make sure that people understand -- Julian has put in an exceptional effort here, and should be recognised for driving this to conclusion.

> I believe we should continue this work with other header fields, and a quite obvious candidate would be "Content-Type", which incidentally has two related HTML WG issues (<http://www.w3.org/html/wg/tracker/issues/125> and <http://www.w3.org/html/wg/tracker/issues/126>).

Please do file bugs as appropriate.


Mark Nottingham   http://www.mnot.net/
Received on Tuesday, 9 November 2010 01:54:06 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 11:10:55 UTC