Re: ERT techniques in UA guidelines?

At 10:22 PM 2000-04-09 -0700, Dick Brown wrote:
>I have an action item from the ER group:
>>> Action DB take issue to UA - should they be considering ERT techniques in
>their guidelines? have they already? include in techniques document?
>I don't have strong feelings that User Agents should employ ERT techniques,
>but when the issue came up we agreed UA folks might be interested in hearing
>about it.


The header guessing algorithm from HTML 4 tables is a repair technique that
I believe the UA group has endorsed.

I wonder what the group's position would be on what a User Agent should do
on processing a stylesheet that contains !important clauses and clearly
identifies itself as CSS1.  Should the User Agent 'repair' this to let the
user's !important rule?

As we move forward to more general flavors of XML, I believe that we will
want to move User Agent functionality to being triggered more by structure
patterns and less by individual marks.

So the idea that User Agents may properly be engaging in 'repair' is not to
be summarily dismissed, I hope.  

On the other hand, the User Agent Guidelines are in a very stable phase.
Unless someone is aware of a repair that the guidelines have missed, that
ERT has designed, and that is a serious flaw not to include, the action to
take right now would seem to be to say that repair techniques may be cited
in the User Agent Techniques in future and move on.

As David Poehlman points out, most of the 'repair' agenda was left behind
with the decision to focus on the minimum requirements for the base
platform and defer consideration of AT techniques.


>Gregory has some thoughts below.

>Dick Brown
>Program Manager, Web Accessibility
>Microsoft Corp.
>-----Original Message-----
>From: 	Gregory J. Rosmaita [] 
>Sent:	Monday, April 03, 2000 11:09 AM
>To:	Evaluation & Repair Interest Group
>Cc:	Jon Gunderson; Ian Jacobs
>Subject:	UA as an ERT tool (was Re: minutes from today's meeting)
>aloha, y'all!
>in the minutes from this morning's telecon, it was stated:
>DB you can consider UA's testing tools. in the future they may be doing
>LK the ideal: techniques part of authoring tools and browsers.  WL can we
>suggest something to UA for UA's to perform validity checks for the user?
>LK i don't think that has been part of the UA philosophy.
>Action DB take issue to UA - should they be considering ERT techniques in 
>their guidelines? have they already? include in techniques document?
>the User Agent Working group has consistently taken the stance that User 
>Agents are _not_ expected to fix invalid or inaccessible HTML-that 
>burden has been placed squarely upon the author's shoulders, and the Web 
>Content, Authoring Tool, and ERT documents are all intended to ensure that 
>the content that is exposed to the user via a user agent is (a) valid and 
>(b) accessible
>that being said, a simple warning, such as that which Lynx issues when it 
>encounters invalid HTML
>Warning! Bad HTML! Use -trace to diagnose
>would be a beneficial feature for a User Agent, but as far as expecting a
>user agent to perform repairs or even an analysis of the page, isn't
>something that the User Agent Guidelines currently address, nor is it
>something that the WG expects as part of the functionality of a user agent
>qua user agent...
>a user agent that also functions as an editing tool, such as Amaya, however,
>would be required to perform evaluation and repair, but only in editing
>finally, Lynx can-by using the -trace command line switch-be used as 
>a diagnostic tool-for more info, consult:
>ACCOUNTABILITY, n.  The mother of caution.
>* Ambrose Bierce, _The Devil's Dictionary_
>Gregory J. Rosmaita      <>
>Camera Obscura           <>
>VICUG NYC                <>
>Read 'Em & Speak         <>

Received on Monday, 10 April 2000 09:16:31 UTC