RE: ISSUE-133 -- How do our definition of Web Page and the Robustiness section interact?

I was in the process of writing this up as a new "good issue" when I
realized it would not be on the agenda for the F2F then. So I will
respond a bit in email and also dial in for a bit (as long as I can stay
awake). Unfortunately I think a lot of text would need to be modified to
fix this issue. Maybe more than we have time for? 

 

Looking for sections where plug-in conformance is an issue I came up
with the following:

 

Section 4.1

I proposed a text change here earlier. I think it is still valid.

 

Section 5.1

Propose adding at the end --

"User agent plug-ins MUST conform to the same requirements as the web
user agent for this section."

 

Section 5.5.1 

Propose adding at the end -- 

"User agent plug-ins MUST either conform to this specification or defer
handling to the web user agent."

 

Section 5.5.3 

Propose adding at the end -- 

"User agent plug-ins MUST either conform to this specification or defer
handling to the web user agent."

 

Section 6.1 

Not sure what to do here -- what happens when the majority of the
visible content is not from the web site? User agent plug-ins have no
general way of signaling the identity of the web site they are being
retrieved from. Maybe punt on this?

 

Section 6.1.1

Propose adding at the end -- 

"User agent plug-ins MUST NOT obscure or degrade the rendering of the
identity signal without specific prior user interaction" 

 

Section 6.3

Propose adding at the end -- 

"User agent plug-ins MUST NOT obscure or degrade the rendering of the
TLS indicator without specific prior user interaction" 

 

Section 6.4

Propose adding at the end -- 

"User agent plug-ins MUST NOT obscure or degrade the rendering by the
Web user agent of error signaling of class Warning/Caution or higher.
User agent plug-ins MUST NOT obscure rendering by the Web user agent of
error signaling of class Notification, without specific prior user
interaction. User agent plug-ins should conform to this specification
for all error signaling specific to interaction with the plug-in." 

 

Section 7.1.2

Propose adding at the end -- 

"User agent plug-ins MUST NOT obscure or degrade the rendering of the
user agent's security chrome, without specific prior user interaction" 

 

Section 7.3 

Propose adding at the end --

"User agent plug-ins MUST conform to this specification."

 

Section 7.4 

Propose adding at the end --

"User agent plug-ins MUST conform to this specification."

 

That's all I have for now. I will be online in a few. I will write up a
new issue if it is deemed worthy.

 

Joe

 

________________________________

From: Mary_Ellen_Zurko@notesdev.ibm.com
[mailto:Mary_Ellen_Zurko@notesdev.ibm.com] 
Sent: Wednesday, May 07, 2008 5:22 AM
To: Joe Steele
Cc: public-wsc-wg@w3.org
Subject: Re: ISSUE-133 -- How do our definition of Web Page and the
Robustiness section interact?

 


Hi Joe, 

> but I think we should add a paragraph at the start of the Robustness
> section (7.0) which says something like: 
>   
> Plug-ins or external systems which render Web page content SHOULD 
> follow the Robustness recommendations and SHOULD NOT interfere with 
> any recommendations which are a MUST for the user agent. 

It sounds like you're saying that this specification applies to
"plug-ins and external systems which render Web page content". Or only
the Robustness section does? I believe we'd want all the spec to apply
to "plug-ins" (I'll use that shorthand here), to the extent they made
sense. But not all will make sense (if the main user agent display SCI
in secondary chrome, there's certainly no need for any plug-ins to). But
if a plug-in diddled with the identity display, for example, we'd want
the combination to still conform (which is what your SHOULD NOT) is
getting at. Does that make sense? The definition we've been using for
Web User Agent, assuming a logical and, does not include most (all?)
plug-ins: 

[Definition: A Web User Agent is any software that retrieves and
presents Web content for users.] 

And we wouldn't want to say a plug-in was conformant if it didn't follow
a MUST in Robustness, for example. So overlaying a SHOULD on top of that
might water it down. 

So is this a place for the conformance labels section I've been asking
to strip out because I don't understand it (Thomas)? 

I appreciate your attempt to deal with plug-ins in one compact place.
I'm just not sure it can work. Even changing the SHOULDs to MUSTs.
Although let me take a stab and see what folks think. 

Add to the Overview (section 4.1): 

Plug-ins or external systems which render Web page content conform to
this specification if they conform to all user agent requirements that
apply to the rendering of content (e.g. [ref Robustness]) and do not
degrade the conformance of the user agent to any other recommendations. 

No, I still think not. We've got to list them somewhere. So that we can
be sure, during conformance testing, that we know what we're talking
about. 

I think this is worthy of its very own issue. Want to create one? We
have guidance on what a "good" issue looks like in our wiki: 
http://www.w3.org/2006/WSC/wiki/WriteGoodIssue
<http://www.w3.org/2006/WSC/wiki/WriteGoodIssue>  

> Suggestion -- 
> This specification makes no specific assumption about the content 
> with which the user interacts, except for one: There is a top-level 
> Web page that is identified by a URI [RFC3986]. The Web page content
> is served as part of a Web interaction. The page's content can be 
> interpreted and rendered by some combination of the user agent and 
> plug-ins to the user agent and external systems triggered by the user
agent. 
> The page's behavior might be further determined by scripting, 
> stylesheets, and other mechanisms. 

I don't have an opinion either way on this, even after going back and
reading the existing text. Anyone else?