Re: [http-state] Goals for the UA conformance section (was Re: Updated draft)

Ian Hickson <ian@hixie.ch> Wed, 19 August 2009 06:21 UTC

Date: Wed, 19 Aug 2009 06:20:23 +0000
From: Ian Hickson <ian@hixie.ch>
To: Daniel Stenberg <daniel@haxx.se>
In-Reply-To: <alpine.DEB.2.00.0908190810500.4004@yvahk2.pbagnpgbe.fr>
Message-ID: <Pine.LNX.4.62.0908190619030.31215@hixie.dreamhostps.com>
References: <7789133a0908180027n644d8abcxc2327b6f2f6a475f@mail.gmail.com> <alpine.DEB.2.00.0908180929540.17475@yvahk2.pbagnpgbe.fr> <Pine.LNX.4.62.0908190553060.31215@hixie.dreamhostps.com> <alpine.DEB.2.00.0908190810500.4004@yvahk2.pbagnpgbe.fr>
Content-Language: en-GB-hixie
Content-Style-Type: text/css
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset="US-ASCII"
Cc: http-state <http-state@ietf.org>
Subject: Re: [http-state] Goals for the UA conformance section (was Re: Updated draft)
Precedence: list

On Wed, 19 Aug 2009, Daniel Stenberg wrote:
> On Wed, 19 Aug 2009, Ian Hickson wrote:
> > 
> > In general, specs should be specific about what the output should be. 
> > We shouldn't leave things undefined or up to the implementation if it 
> > affects other implementations (such as servers), because otherwise we 
> > will find authors end up relying on the market leader's behaviour. So 
> > we have to define the sort order whatever we do. It makes sense to 
> > pick the sort order used by the browsers, IMHO.
> 
> But again you're not focusing on how things ACTUALLY are already right 
> now. You're talking about how we should document (future) things to be.
> 
> If we're documenting how cookies are used in the wild today, we cannot 
> claim that 99.9% are sorted in path lenght order. That's a fact.
> 
> If we're documenting how cookies should work tomorrow with improved 
> interoperability and a minimum of changes, then sorting cookies by path 
> length seems like a good bet since a large amount of clients clearly 
> already do it.

We're documenting (or should be documenting, IMHO) what all 
implementations are going to converge on. That means describing what they 
do when they do the same thing, and considering differences to be bugs for 
which we are to find solutions that are acceptable to implementors.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

[http-state] Goals for the UA conformance section… Adam Barth
Re: [http-state] Goals for the UA conformance sec… Daniel Stenberg
Re: [http-state] Goals for the UA conformance sec… Anne van Kesteren
Re: [http-state] Goals for the UA conformance sec… Daniel Stenberg
Re: [http-state] Goals for the UA conformance sec… Julian Reschke
Re: [http-state] Goals for the UA conformance sec… Anne van Kesteren
Re: [http-state] Goals for the UA conformance sec… Dan Winship
Re: [http-state] Goals for the UA conformance sec… Julian Reschke
Re: [http-state] Goals for the UA conformance sec… Anne van Kesteren
Re: [http-state] Goals for the UA conformance sec… Adam Barth
Re: [http-state] Goals for the UA conformance sec… Bil Corry
[http-state] Charter feedback Anne van Kesteren
Re: [http-state] Goals for the UA conformance sec… Ian Hickson
Re: [http-state] Goals for the UA conformance sec… Ian Hickson
Re: [http-state] Goals for the UA conformance sec… Daniel Stenberg
Re: [http-state] Goals for the UA conformance sec… Dan Winship
Re: [http-state] Goals for the UA conformance sec… Daniel Stenberg
Re: [http-state] Goals for the UA conformance sec… Adam Barth
Re: [http-state] Goals for the UA conformance sec… Dan Winship