W3C

- DRAFT -

User Agent Accessibility Guidelines Working Group Teleconference

25 Apr 2013

See also: IRC log

Attendees

Present
Eric, Greg_Lowney, Jan, Jeanne, KFORD, Kim_Patch
Regrets
Jim, Simon
Chair
kelly
Scribe
KimPatch

Contents


<trackbot> Date: 25 April 2013

survey - finish #10 and partial conformance https://www.w3.org/2002/09/wbs/36791/20130412/

Kelly: let's start with the last one – limited conformance for extensions

<kford> Survey:

<kford> https://www.w3.org/2002/09/wbs/36791/20130412/results

Greg: last thing in chat log – move 9 to 7

Kelly: the last one

Eric: one basic concern is that it's not clear what this is intended to cover. I see two possible major purposes of this section. One is to cover software that is not actually a full user agent. Another possibility is intended to address software that is not quite reach the level a, meaning all the applicable level a success criteria and therefore is below level a, or some combination of the two.
... another concern is it makes reference to a main user agent, and we don't have a definition of that. There's a bunch of issues related to that section and I felt, well given the necessity of nailing down other key pieces of this document I thought it may be better to spend our time on other things rather than to try and carve out something that would raise a bunch of new issues.
... so for that reason I disagreed with the proposal.

Jennne: so you're proposing that we not consider extensions

Eric: can't we address an extension through the regular conformance criteria – if that is indeed a user agent why not use the criteria were already developing or trying to wrap up rather than develop a whole new section that has a lot of issues you have to resolve

Jeanne: what do other people think – should we include extensions? They are not user agents themselves, they are just modifying something and passing it to the user agent

Jan: I think the basic structure is solid, if you wanted to call out an extension for playing nicely in the UAAG environment it would need to basically meet those success criteria that applied to it and not interfere with other parts of the ecosystem

Jeanne: back in the days when we were originally talking about this,reason is to be able to include extensions like mouseless browsing.

Jan: you tell us what you meet, and that you don't interfere with user agent meeting other SC's

Eric: recall that a user agent relies on a platform to do some of its work – to fulfill its functionality such as rendering content to users. So is the possible that an extension could be viewed as a user agent in a situation where it relies upon a browser as part of its platform. In other words a browser doesn't necessarily have to be the user agent that is the subject of the claim, rather

Jan: I think we've taken that into account – we have some other terms for that. Here we were really talking about something as simple as mouseless browsing which is a very specific thing, and if it were treated as a user agent it would fail

Eric: would fail if it were connected to a user agent?

Jan: but that's a ton of work – you're saying if you want to make a claim you also have to test all of Firefox, which is a big deal. Other option is to claim as an extension, you basically say we do this, we have documentation and we don't brake – we support keyboard accessibility of menu or the menu item that we had or something like that, and we guarantee that we don't break any of the...
... other essays that Firefox needs to

Eric: do you feel that type of conformance is to be documented in this document or needs to be part of some other document that shows that UAAG can still be helpful to you if you are concerned about an extension that's not really in of itself

Jan: I think it encourages more software to get a foot on the bottom of the ladder. We have a dozen full on browsers in the world, but if we have other extensions joining in and making these related extension claims it grows the ecosystem a bit more. But if people feel strongly that it shouldn't be included I am fine

Jeanne: it also increases the chances that browsers can pass – with an extension – that's an advantage

Greg: I lean toward leaving it in for all the reasons that Jan discussed. If you consider we want people to be able to add commercial add-ins, makes sense. Splitting into another document would require a lot of bureaucratic wrangling on the part of our working group to create something with teeth as opposed to an advisory thing. So if we want to have it possible to do, including it in the...
... main document is the easiest way for us to do that.

Eric: in the example of mouseless browsing, is there anything such as a platform for it? Does it have a platform – does it require… I'm trying to think through why doesn't fit within the main section and is it just a matter that it's really burdensome on the developer to try and document the nature of the platform because – let's say the platform is a browser, then they've got to go...
... through...
... and document each piece of the browser, or – I'm just trying to think why it would be burdensome to require somebody a developer of an extension to document

Jan: it's burdensome on them obviously if you just do one little thing it's burdensome to have to make claims about the whole browser you are working with. In looking at it on the Firefox side, maybe there's more than one extension: ABC and D. You don't want to be tied to making a claim just with A. We leave it to them, it can be done. We've had that whole conversation before.

Eric: can we clarify that this is for software that is not actually a user agent or that is – has very limited functionality

Greg: that is what the first sentence says – this option may be used for a plug-in with limited functionality

Eric: do we have a definition for extension? Looking at the language – extension or plug-in with limited functionality. Does it have to be an extension or does it have to be a plug-in – anything that limited functionality?

Greg: can you give us an example of something that would not be a user agent or a plug-in?

Eric: what if somebody wanted to evaluate a limited function web browser

Greg: that would be user agent

<Jan> From the draft: user agent extension (add-in)

<Jan> Software installed into a user agent that adds one or more additional features that modify the behavior of the user agent. Two common capabilities for user agent extensions are the ability to *modify the content* before the user agent renders it (e.g., to add highlights if certain types of alternative content are present) and to *modify the user agent's own user interface * (e.g. add a...

<Jan> ...headings view).

<Jan> plug-in [ATAG 2.0]

<Jan> A plug-in is a program that runs as part of the user agent and that is not part of content. Users generally choose to include or exclude plug-ins from their user agents.

Eric: well, how limited doesn't have to be – what's the dividing line between things that have to be evaluated in the main section versus something that can offer this. This is a very abbreviated thing. It's something that's – are people going to abuse it by trying to evaluate software that is more limited than we really intended?

Jeanne: I think it's important to keep in mind that we are probably the only people that will ever use this section. Companies typically do not like to file conformance claims because it opens them to lawsuits. So I think it's important to realize that were the ones who are going to use this – we are going to use this to get this document for our recommendations. So if we spend a lot of time...
... and edge cases we're not going to get it done.

Eric: I guess what I would like – if we have this section what I would like to see is does it have to be something that meets the criteria of a user agent or can it be – it ought to refer to the class of software that it applies to ought to be defined and we ought to tell explicitly how it differs from the user agent. Because what I hear you saying is the thing that this is for are not...
... actually user agents in the sense you've defined it.

Jan: they are parts of user agents, but they add functionality

Greg: note that we do have a definition

Jan: I already pasted in

Greg: the issue on the table is whether to keep the section were deleted, correct?

Jan: also issue of what it should say – I think we should quickly come to a conclusion of what it should say, and then decide later if we should keep the section. We don't want it to be a section where we don't even know what the wording is going to be

Kelly: so we should resolve the content

Jan: something around the general principle of let us know what you think is applicable to you as an extension, certify that you met it, certify that you don't prevent any other parts of the user agent that you are plugging into from eating what it's supposed to meet, and there you go. Once we get that text firmed up we can put an editor's note into the actual document signed that will come...
... back later and decide whether we want this in the final draft.

Eric: it seems like this section is for something that is part of the user agent, but is not – I suppose you could have a part of the user agent that is also user agent itself, but what I hear is if it were a user agent itself it would be evaluated in the main section. So if it's a part of a user agent that is not a user agent itself it's a – and your prime example would be an add-in or...
... plug...
... in, so that would be e.g. add-on, plug-in. That would help clarify what is being routed to the main section and what belongs in this section. It's a part of the user agent that is not – that doesn't have the capabilities that are defined as being part of a user agent. It doesn't satisfy all those characteristics, but it meets some of them and therefore a subset of the success criteria...
... should apply so your responsibility is to tell which you think apply and also to certify that it doesn't – prevent

Jeanne: prevent the main user agent from eating the success criteria

Eric: okay, so – based on the way I just defined the purpose of this one way – if you say that this applies to a part, that the software is part of the user agent then you need to specify the user agent of which it is a part, right?

Greg: for any claim you have you have to define what the user agent is. That could be further clarified by giving one of the in-line examples, where we describe a conformance claim including the user agent

Eric: I like what Greg has said but instead of saying user agent – were kind of turning this thing on its head as a partial conformance section – so to clarify it, would it make sense to say that the platform for the subject of this partial conformance claim has a more extensive user agent type capabilities? I guess you could say it's a user agent, but literally we're sort of –...
... essentially...
... the reason I was tripping over user agent is the paradigm that dominates in the document my assumption is the user agent is the core subject of the claim. It's supplemented by its platform. And so this kind of turns it on its head if we also call the platform of a subject of partial conformance a user agent.

Greg: I think we've already discussed that the platform can be a user agent for example for a plug-in

Kelly: what are the consequences if we just delete this section

Jeanne: big consequences for conformance

Eric: it doesn't really help – if you are relying on this – we have the wrong requirements, in other words every success criterion should be found in at least one user agent – something that meets our definition of user agent

Jeanne: It's really late to start this argument. Going through now is starting to eliminate all the success criteria that are currently implemented in the major browsers would be a vast change to the document. It's certainly not something I'm going to support a mustard overwhelming group agreement against it.

Eric: are you confident the W3C would allow us to count things that are instantiated in this partial conformance section

Jeanne: if it's written into conformance, yes, I can reasonably say that if we've written it so that they can conform than we can use it

Eric: if this is an important section to have I don't want to open an issue that doesn't have to be opened. Do we need to tweak our definition of user agent to say that it does – it also includes things that may do a subset of retrieve, render, facilitate user interaction. Earlier we said it had to meet all three

Jan: facilitate user interaction is very broad. Some of them are pushing the envelope of what is currently in the market. mouseless browsing is a good example. We don't want to push so far that it's a dreamland. But mouseless browsing is not in the out-of-the-box browser that you download, but available as an extension which is easily excessive will to anybody who wants to use browsers....
... There are several implementations of the ideas. And just as these are plugged into the browser we kind of see our document the same way – it's plugged into the requirements. You can make a conformance claim based on Firefox plus mouseless browsing, particular mouseless browsing 2.35 or whatever. Or from mouseless browsing perspective where it's just that small part, they don't have a...
... claim for the whole ecosystem they can just do thIndia Romeo part. it's called out as an extension. I don't think there's a danger of Firefox submitting a claim and calling it an extension.

Kelly: now we're up to 50 minutes on this

Greg: I support Jan suggestion that we move on to talking about it. Assuming that we want to keep it how do we want to word it.

Kelly: Can we get to a point on this call where we can call this done? This is one of the last things content wise that we have to work on in this document. It would be nice to be done today with this

Greg: I support the basic idea – two content wording suggestions. Seems to have dropped the part about not preventing the host user agent from conforming with any UAAG success criteria.

Jeanne: I'm good with that – we probably need to check with Jim and make sure he is

Greg: a technical definition would be something like the hosting user agent does not fail any success criteria when this add-in or extension is installed – that it passes without the add-in or extension is installed
... that's the point – not pass every SC, but not fail any because this extension is installed. Something like the hosting user agent – would that be another SC or is it just a sentence in this – can we make it a requirement without having SC or certifying it can be verified

Jan: I think we should steer away from that not interfering with other functionalities because there are so many other functionalities that can be plugged in

<Greg> "The hosting user agent does not fail any success criteria when the add-in or extension is installed that it passes when the add-in or extension is not installed."

<Greg> Or "Installing the add-in does not cause the hosting user agent to fail any success criteria that it would otherwise pass."

Eric: do we need to make reference to hosting user agent at its level. If you want your extension to conform of the AA level then would it be the case that your hosting user agent – the success criteria that you are referencing are based on a AA conformance level for your hosting user agent? Should there be any kind of matching? I'm just trying to think of a way of being a little bit more...
... specific about which success criteria you have to certify that it doesn't interfere with?

Greg: pasted in language

General agreement with Greg's language

<Greg> Full paragraph: This option may be used for an extension or plug-in with limited functionality that wishes to claim UAAG 2.0 conformance. An extension or plugin can claim conformance for a specific success criterion or a narrow range of success criteria, as long as installing the add-in does not cause the hosting user agent to fail any success criteria that it would otherwise pass."

<Greg> "This option may be used for an extension or plug-in with limited functionality that wishes to claim UAAG 2.0 conformance. An extension or plugin can claim conformance for a specific success criterion or a narrow range of success criteria. The add-in must not cause the hosting user agent to fail any success criteria that it would otherwise pass."

<Greg> "This option may be used for an extension or plug-in with limited functionality that wishes to claim UAAG 2.0 conformance. An extension or plugin can claim conformance for a specific success criterion or a narrow range of success criteria. All other success criteria may be denoted as Not Applicable. The add-in must not cause the hosting user agent to fail any success criteria that it would...

<Greg> ...otherwise pass."

<Jan> This option may be used for a user agent extension or plug-in with limited functionality that wishes to claim UAAG 2.0 conformance. An extension or plugin can claim conformance for a specific success criterion or a narrow range of success criteria. All other success criteria may be denoted as Not Applicable. The add-in must not cause the hosting user agent to fail any success criteria that...

<Jan> ...it would otherwise pass.

Jeanne: I like it

Eric: claimant specified success criteria? Not success criteria we enumerate for them, but something they enumerate for us as part of the claim?

Jeanne: is that not clear

<jeanne> This option may be used for a user agent extension or plug-in with limited functionality that wishes to claim UAAG 2.0 conformance. An extension or plugin can claim conformance for a specific success criterion or a narrow range of success criteria stated in the claim. All other success criteria may be denoted as Not Applicable. The add-in must not cause the hosting user agent to fail any success

<jeanne> criteria that...

<jeanne> <Jan> ...it would otherwise pass.

Eric: it's not clear – who is specifying it is not stated. Because the whole rest of the document is about us specifying the criteria they need to meet, now they're specifyingwhich success criteria that need to meet

<Jan> This option may be used for a user agent extension or plug-in with limited functionality that wishes to claim UAAG 2.0 conformance. An extension or plugin can claim conformance for a specific success criterion or a narrow range of success criteria. All other success criteria may be denoted as Not Applicable. The add-in must not cause the combined user agent (hosting user agent plus installed...

<Jan> ...extension or plug-in) to fail any success criteria that the hosting user agent would otherwise pass.

Eric: it's got to be implicit that the user agent has a platform

Jan: yes, remember that this may even be embedded – we maybe five or six layers down here

<Jan> This option may be used for a user agent extension or plug-in with limited functionality that wishes to claim UAAG 2.0 conformance. An extension or plugin can claim conformance for a specific success criterion or a narrow range of success criteria a stated in the claim. All other success criteria may be denoted as Not Applicable. The add-in must not cause the combined user agent (hosting...

Eric: in terms of style, software wishes to claim conformance, it's rather the claimant that wishes to claim conformance, stylistic or editorial thing

<Jan> ...user agent plus installed extension or plug-in) to fail any success criteria that the hosting user agent would otherwise pass.

Greg: I do have one more substantial playing. The second sentence: I just want to make sure that were not allowing them to claim that they pass when they only – when they actually fail a bunch of success criteria. Does this open the door for them to just say not applicable – are we giving them too much leeway to decide what to categorize not applicable?

Jan: doesn't the last sentence take care of it – you have to imagine yourself combined with the user agent and you can't fail anything that it would otherwise pass

Greg: here's an exception to that – agent that does not render time-based media – doesn't show media. Then you install plug-in which adds the ability to play movies. So before the user agent wanted its own conformance claim said NA two all the things related to time-based media. The plug-in should not be able to say well, we don't do captions that's okay were just going to do NA –...
... we're not...
... forcing our host user agent from past to fail it was NA and it still NA

Jan: you can't do that, now they do time-based media, and so they fail

Greg: if we have examples to clarify it for people it will cut down on any misinterpretations

Jan: yes – examples in implementing

Greg: I agree

Making sure we have addressed everything you put in survey

Kelly: objections?

No objections

don't want to collect a list of all technologies that a browser renders (related to combining conformance items 7&8)

Jeanne: concerned about listing included and excluded technologies

Jan: those testing decide what should be included

Eric: my understanding up until now has been that both included and excluded technologies were both within the subset of the content technologies that were rendered, so it was my understanding was that it was a comprehensive list of technologies rendered

Jan: you're right – so whatever is not in the included list is in the excluded list

Jeanne: remember, we are the ones who are going to be doing this

Greg: if we were to do a claim for Firefox and they can do math ML – would it be okay for us to do a claim the completely ignores the fact that it can do math ML?

Jan: yes

Jeanne: why put extra work on ourselves?

Greg: it does open us up to browsers claiming conformance, but until you read excluded technologies you find out that they're not doing it for HTML 5…

Jan: how long would it be take for something like that to be dug up and put all over the web

Eric: I guess we are talking about whether the sum of the included and excluded technologies is a comprehensive list are not

Jeanne: we do not want to make this a comprehensive list

Eric: up until we discussed this last week my preference was not to make it a comprehensive list
... even though I was in favor of an approach that sounds closer to what Jeanne is saying we did decide to go a different way. But I'm willing to consider a non-comprehensive list.

Greg: I'm willing to go with it as well. Companies are willing to go through amazing gymnastics and used tortured logic. But I don't see it as an advantage for companies filling in the excluded list at all if – sounds like they are not required to anymore –so they can just leave it blank. Exactly the same as not addressing it in either list. I can live with that of people want to go that...
... direction

<Jan> http://en.wikipedia.org/wiki/Firefox#Standards

Jan: I would like to keep it, if we want to soften the language – the main formats – something like that. It was on the excluded list it would be easy to spot for on the excluded list.

Eric: pursuing this route I would kind of go back to what my original proposal – enumerating a set of at least one rendered content technologies is mandatory but enumerating one or more excluded technologies would be optional

Jan: I prefer it not to be completely optional because you're right, Greg's example you don't want to duck out of, for example HTML for Firefox. It tails off to some pretty unimportant stuff so can we put it in there and just soften it?

Kelly: where does this leave us?

Jan: if it's conquered to be one of the other, fine

Greg: my preference would be to keep it if we could, but I'm willing to drop it if we have to

Jeanne: what I suggest is we work on softening that language during editing and run it by the group next week

<jeanne> ACTION: jeanne to soften the wording of Excluded/ Included Technologies so that we are not requiring a comprehensive list. [recorded in http://www.w3.org/2013/04/25-ua-minutes.html#action01]

<trackbot> Created ACTION-825 - Soften the wording of Excluded/ Included Technologies so that we are not requiring a comprehensive list. [on Jeanne F Spellman - due 2013-05-02].

Eric: I have a question – is it our approach that if there is a constraint imposed by the platform that the user agent or the claimant can declare a success criteria that is impossible due to the limitation of the agent is not applicable?

Greg: I have three or four sub categories of not applicable and one of those is because the platform limitation

Eric: do we include that language in our document right now

Greg: I'm not sure

I don't consider the whole nonconformance claim to be part of the partial answer

Greg: I don't consider the whole nonconformance claim to be part of the partial answer. List

Eric: shorter list – one of them is if a success criteria higher performance than is claimed then that success criteria is not applicable. That could be at the head of the list

<jeanne> ACTION: item to jeanne to find Greg's email and include the list of reasons of why a success criterion can be declared not applicable. [recorded in http://www.w3.org/2013/04/25-ua-minutes.html#action02]

<trackbot> Error finding 'item'. You can review and register nicknames at <http://www.w3.org/WAI/UA/tracker/users>.

Jeanne: except that sometimes AA is something different

Greg: be careful, there's a very limited set of rationale for why you claim something is not applicable. I don't know of anything in level A that becomes not applicable because of something you're doing in AA

Eric: only if the success criteria of the not applicable exceeds the level of conformance

Greg: if they're only going for level a all the ones that were AA or AAA could be marked with something else such as not claimed to distinguish it from non-applicable. Because I'd like to keep none applicable for something that meets one of those three or four very specific things and you know what the answer is as opposed to were not even dealing with this becausewe are not bothering to...
... check the AAA's.

Jan: I agree, it's just blank

Jeanne: that's the goal – only give people the tests for the level they need

Greg: people may have their own forms

Jeanne: very few people like to fill a conformance claims because their lawyers don't like it

Eric: if the claims are our job it seems backwards. Hopefully there's motivation for them to make the claim. I can't see it as being up to us

Jeanne: the only reason is up to us is we have to prove

Kelly: we are at the stopping point. We pulled down a couple thorny issues. Two things for next time, how do we shorten our timeline, do things in parallel. We also have to think about her levels.

Summary of Action Items

[NEW] ACTION: item to jeanne to find Greg's email and include the list of reasons of why a success criterion can be declared not applicable. [recorded in http://www.w3.org/2013/04/25-ua-minutes.html#action02]
[NEW] ACTION: jeanne to soften the wording of Excluded/ Included Technologies so that we are not requiring a comprehensive list. [recorded in http://www.w3.org/2013/04/25-ua-minutes.html#action01]
 
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.138 (CVS log)
$Date: 2013-04-25 18:36:49 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.138  of Date: 2013-04-25 13:59:11  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

No ScribeNick specified.  Guessing ScribeNick: KimPatch
Inferring Scribes: KimPatch
Default Present: Jeanne, +1.609.734.aaaa, Jan, Kim_Patch, Greg_Lowney, KFORD
Present: Eric Greg_Lowney Jan Jeanne KFORD Kim_Patch
Regrets: Jim Simon
Found Date: 25 Apr 2013
Guessing minutes URL: http://www.w3.org/2013/04/25-ua-minutes.html
People with action items: item jeanne

[End of scribe.perl diagnostic output]