User Agent Accessibility Guidelines Working Group Teleconference

07 Feb 2013

See also: IRC log


Jeanne, EricHansen, Jan, Greg_Lowney, Jim_Allan, Kim_Patch, sharper, kford
JimAllan, KellyFord


<trackbot> Date: 07 February 2013

<kford> http://lists.w3.org/Archives/Public/w3c-wai-ua/2013JanMar/0026.html

<kford> http://lists.w3.org/Archives/Public/w3c-wai-ua/2013JanMar/0015.htm

<kford> --


<allanj> tomorrow- Friday there will be a meeting at MIT to find extensions and plugings to meet SC in UAAG

will be a call in number as well for tomorrow's session

<allanj> there will be a call in number for those who want

<allanj> will be documenting results in the wiki

conformance use cases for conformance scenarios (extensions, mobile apps, etc)

Jim: the proposal that I sent to the list

<allanj> https://www.w3.org/2002/09/wbs/36791/20130205/results

Jim: Greg's response – why was compliant left out, and if you get a single NA is it full compliance?

Jan: . I agree with your objection there – if it's not applicable accounts as a pass for conformance

agreement all around

Greg: so wording change needed
... it does not discuss NA, and it would need to

Jim: so for the full UA compliance, which is the bit we were talking about of the top it just says you have to meet all of them – so there might be some NA's in there – some clause at the end

Greg: why is this wording so different from the need for partial compliance which says you have to list everything, as opposed to just a blanket statement. I think it's too easy to make a statement like that rather than actually – providing the form that lists all the SC and for every SC fills in one of the three categories and subcodes for compliance
... either needs to repeat the same wording for both categories – full conformance or partial conformance or extract that out and make it part of the overall claim, but it shouldn't be for just one

<Jan> http://www.w3.org/TR/ATAG20/#conf-applic

Jan: may be text from atag that's useful

<Jan> (Some ATAG text re: N/A)

Jeanne: making a little more difficult for people to apply for conformance when one of the goals we should be going for is if it conforms, we make it easy for them to do the forms

Greg: I think everyone would agree that in order to apply for conformance you have to test everything – each of the SC. So if they have to do that – turning the checklist – I'm afraid that if they turn in a paragraph that says we apply for everything it's a lot easier for them to assume they do without making sure they have checked off every single SC

Jeanne: keep in mind that the major use of this conformance statement is going to be by us – we are going to be doing these conformance statements for the products that we are testing
... very few people have gone through the process of filing a claim for WCAG. Conform, but don't file.
... so unconcerned that were making this process more complicated for us and was likely for other people to do. I want to have good information and to make this a good product, but I think we need to make conformance less process

Jim: first time around – we found to implementation scattered across 30 or 40 products – none could ever be combined as a unit to make a fully accessible anything, but we did show that everything we asked for was done somewhere, just not together at the same time

Greg: but if you really think that no one out there is going to file a conformance claim why should be spent a lot of time developing the conformance section of this document, or should we leave it freeform because it won't really matter

Jeanne: thinking of the audience – it certainly does matter to us improving that our document does have real world implementation, but I don't think the audience for the conformance purchasing agents and users, consumers. I think the conformance matters to the legal departments of companies that have accessibility issues if they want to make sure they comply.

Greg: if were the only ones recording conformance claims – for our purposes with release would be simply a list of examples where each SC applies in a product, different from a formal conformance claim

Jeanne: the conformance claim matters to the company that's filing it. And it matters to us to prove that we have conformance implementations in the real world.

Greg: but the latter could be done simply as a document which lists examples – as Jim said features scattered across different products, there are at least two browsers implementing each SC, as opposed to testing a browser and listing every SC, which is what I see as a conformance claim

<allanj> scribe: KimPatch

Greg: I can see us doing it as an exercise to make sure we have the form well-designed, but we probably won't do it for a lot of browsers

<allanj> KP: make the form simple. so it is easy for people to do.

Greg: what I see is the real unsolvable problem is that the process of testing is extremely lengthy and time-consuming because there are so many different features in most browsers that in an ideal world you would be going through making sure that each one of those different views, each one of those pop-ups or whatever all conform with the applicable SC. And I know from experience to do any...
... test against every feature in a product will be a very lengthy process and very expensive for companies to do
... not that we're on the wrong track or anything – just keep that in mind

Jim: we are going to leave the testing process aside, and we are working on the conformance claim. So if we put in – fleshed out the wording a little bit better so we explained that if there are NAs that they still count, and if they're not compliant, then it's partial conformance and that's the next one – is that good if I fix that

Greg: yes – a passing experiences either it applies, or not applicable

Jim: so I will fix that language

Greg: that was the second comment – the first one

Jim: cut-and-paste error – I'll fix that also

Greg: third comment, extension isn't defined

Jan: extension needs to be defined

Jim: link to the glossary?

Jan: yes, different terms extensions, plug-ins. And then there is the embedded user agent which is sometimes called plug-ins as well

Greg: perhaps we should create one term for anything that's embedded and then subcategories for nested user agent

Jan: I would prefer to keep them separate – they seem too different to me

Greg: that's fine. We'll have to note in the definition if we define extension that some products use different terms for it
... so we will keep two terms, one for add-ins, another for – will call plug-ins which have full control over one region

Jan: plug-ins have full responsibility for rendering the content in a certain technology

Greg: given those two terms – both of those terms need to be in this section here

Jan: also when you are making a claim you are making a claim on the basis of a certain technology. Self I made a claim that it meets UAAG for HTML 4, it might just ignore flash, whereas the flash plug-in might want to make a claim about its rendering of flash

Greg: what you're saying is a user agent might never have to cite a plug-in is a requisite for complying?

Jan: they might out-of-the-box make a claim about the rendering of HTML but perhaps a client says we have a lot of flash so how are you going to handle that– if we leave it open to them to add a plug-in.

Or in the context of what they are plugged into

Greg: at the high level of my comment was something anything at the level of a plug-in might assist in a task, maybe not just extensions, should be left open to things other than extensions

Jan: should be left open

<allanj> from Eric:

<allanj> Maximal Accessibility

<allanj> Each claimant is encouraged to define a user agent that addresses the highest conformance level possible (A, AA, or AAA) and satisfies the greatest number of additional success criteria beyond the claimed level. Doing this may necessitate defining the user agent may include multiple software components (e.g., browser, plug-ins, extensions). Thus, a claimant could define a user agent as...

<allanj> ...including browser X with plug-in Y, or could define it as including browser X without any plug-ins.

<Greg> That is, plug-ins, assistive technology, or other tools, not just extensions, may be necessary to bring the user agent into compliance

Eric: those three things maybe with media players might be the sort of things that those additional software components might consist of, but we don't necessarily have to enumerate all of them. I'm agreeing with you that leaving it somewhat open is good
... I guess the key point is the user agent could consist of multiple software components

Greg: I prefer to think of it as the user agent plus some additional things can comply rather than thinking of the user agent consists of the browser plus some additional things – that's just the language difference

<sharper> I can take it

<sharper> Goes back to the define of the user agent - what we count as one

<kford> EH It goes back to the definition of the user agent and the kinds of things we count as a user agent.

Eric: that we allow the client to specify the boundaries of the user agent and that's what gets evaluated

<kford> EH: We should allow the claimant to specify the boundaries of the user agent and that's what gets evaluated.

Eric: I think Greg stated different preference that a user agent is more narrow

<kford> EH: I think GL has said the user agent definition is more narrow. It is the browser and other components that are the subject.

<kford> EH: I think JR made refernece to a total user agent system. The fact that he used that says there are some issues we should resolve.

<kford> EH: What do we call this and what exactly is the subject of the claim.

Jan: it's going to draw in a plug-in, extensions mouseless browsing which it doesn't do, keyboards – those kinds of ATR separate. On the other hand I've heard of certain extensions, mouseless browsing for example as AT.

Eric: just to summarize, really what the subject of the claim is really isn't user agent specific

Greg: most of the time it won't be that if they really want to cite full conformance – I doubt any single product without add-ins will be able to comply with everything

Jim: I think we resolved all the little bits about the extension and the plug-ins – compliant, not compliant and how we go about the second part of the full compliance. I'd like to move on to our partial conformance bit.
... wording of the first part is incorrect comment – which first part

Greg: your first paragraph – the phrase "because of intrinsic limitations of the platform" – I would just take it out… or replace it with…
... All the NA stuff is not going to be under here – it's going to be up above… The way I look at it is for every SC there's one of these responses – either comply, not applicable, for you fail to comply. So the whole partial conformance thing is not any different other than some of SC are not compliant, and that could be for any reason, important that you list with the reasons are

Eric: I agree with Greg's point about these NA pieces not belonging here. There could be any reason – whatever the relevant reasons there are for not satisfying a success criterion that could count as partial conformance. I'm not sure why it makes reference there are two intrinsic limitations of platform. It could be any reason. I think partial conformance – it's saying that some success...
... criterion have been satisfied yet either no level has been met – A AA, AAA – or… Actually just simplest to limit partial conformance of things that do not meet any of the three levels – they are below a

Jan: partial conformance to the two other groups is not that at all – because you could just meet a single A criteria and pass
... WCAG several types of partial conformance – there really is no screen reader in Swahili, there may be one later but for now this is not supported. ATAG reason is not conformance because platform does not have an accessibility API

Eric: so situations such as intrinsic limitations of the platform that could be the scope of partial conformance but I so think I agree with Greg that these different non-applicable categories are something that really belong under full conformance and really I'm questioning whether any of them should show up.

Greg: tell me if this is a correct paraphrase – basically you can claim that you are compliant if you comply with everything that applies to you –
... example, monochrome Kindle. If a person needs something about color they can't use the Kindle at all therefore you don't conform but you're no worse than the rest of us – I'm as accessible for anything for this platform can be and that would be partial conformance, not what the term partial conformance seems to me to be at least, which is I don't comply with everything in a general sense

Jeanne: group yesterday – would like us to use a different word than partial conformance for several reasons. Incomplete conformance should be very concrete like its platform based.

Jan: non-applicable input – that's a pass. HTML subset doesn't include video – nonapplicable pass. Things with a platform if it's not there because the platform doesn't allow you, that's kind of important information. The output type I find troublesome – could be I'm a video player and I don't do captions because I don't do text

Greg: now that I understand – confused by an NA's – could give it a title I'm as compliant as the platform allows. Hardware not software it chooses to use.

Jan: I see a parallel between this and the Swahili screen reader example. That site that's created in Swahili despite the fact that there's no screen reader support for it– why do they do that, they're trying to serve the Swahili community. Same for why they put it on the Kindle – they are trying to serve the Kindle community. Trying to piggyback on the WCAG term.
... maybe we could be even more clear about the term platform and some kind of partial conformance on the basis of hardware device and operating system combination

Greg: that's another discussion
... as for just the partial conformance – partial conformance due to things outside our control is what I would be getting at, as opposed to platform or hardware. That would take care of this whole thing about the Swahili example category as well

<Greg> "partially conformant due to external factors"

Greg: partial conformance due to external factors?

Jeanne: about limited conformance

Greg: doesn't include why – that's not what we mean by our second category

<Greg> I want to avoid using a term which colloquially would apply to failing some SC due to choice, rather than because you have no choice.

Eric: I've been connecting what we call partial conformance to the idea of a lack of control – why can't we fold that in to a term with we've already identified as appropriate rationale – three mentioned in Jan's memo –
... claim level, unrecognizable content, lack of developer control
... can we fold this into this area of lack of developer control? If we have to document lack of developer control, we could say for example that if success criteria is not applicable for lack of developer control and that lack of developer control drives from limitations of the platform than you need to document that fact

Jan: if it's a full pass you can end up saying something ridiculous like the first compliant browser is on a mobile browser that is black and white…
... but if it's partial compliant, what's that, that makes sense

Greg: problem with terminology is anything to claim conformance if it meets one SC. And if it meets every SC except for the fact that there's no screen reader in Swahili – needs to be set up so you can tell the difference

Jim: I don't know that you finish this in the next 15 minutes. I'll make changes based on the comments we got. Time to take a breather on this. We are getting a much clearer picture. I think it will be better next time
... We can go to Eric's proposal.

Eric Hansen review of UAAG2 20 Dec 2012 editor's draft

<allanj> http://lists.w3.org/Archives/Public/w3c-wai-ua/2013JanMar/0015.html

Eric: I thought the number one issue was clarifying – a degree and flexibility the claimant has in clarifying the user agent. I think that was sort of the subject of our discussion today
... second major point was related to having clear criteria about acceptable versus unacceptable rationale for non-applicability – also part of our topics today
... a number of terms that needed some glossary definitions: two different terms for the same thing, older term called collated text transcript, which I find pretty easy to remember what it is, and the other was something like alternative for time-based media. Essentially they are the same thing. That's one example of the need to reconcile and define terms.
... there were a number of instances of linking to a to be created glossary definition. I thought a number of terms probably needs a short definition in context even though you would also link to a definition, just for the reader not having to jump around I think it's helpful to define some of those terms in the body, at least in a summary fashion

Jeanne: those are the things I consider editorial – will take care of them

Eric: come up with a sort of a taxonomy of content. There are a number of times the term content is used and qualified in some way. There are sort of multiple taxonomies for content – default, primary, original, alternative – I've noted a few places where those taxonomies seem to be laid out in some way and I'm wondering if there's some way we can vet those and bring them up higher and...
... earlier in the document so that people can get a mental image of what these different kinds of content are.
... content with or without modifying term – reconciling those
... a number of comments on conformance section, but I'm not sure since we spent so much time in conformance whether to go through that. One general thought about conformance that I'll mention is it seems to me what's emerging is that a conformance claim applies to a user agent system that could consist of just a browser or maybe some other stuff and it always takes place on a platform. I...
... guess it relies on that platform for services. The clear we could be about exactly what a user agent is and what are the minimal components for conformance I think it will help us simplify the conformance scheme and everything else.
... glossary terms that were not used – in a few cases I thought maybe we ought to collaborate with the other documents – WCAG or ATAG and agree

Jim: that's been tried before – difficult
... Eric cemented a word document with comments on the entire document – Jeanne will go through for editorial, if anyone else has commeents, send to list

Greg: What are the big things

Eric: if you search for triple or quadruple Asterix that should take you to a few comments that are may be more important

Jim: I'll go through and send copy with important comments to list
... will also send redone conformance

Eric: I took a crack at working off of Jans document and transmuted into a list of rationales for non-applicability of success criteria – I welcome any reaction to that too

Jim: I'll be incorporating that

Jan: they probably interact – recognized can be at a level below full SC
... example about keystrokes – we have some SCs around the management of keyboard shortcuts, so access keys, but if thedeveloper has used their own keyboard shortcuts, they are unknowable, they won't be recognized

Eric: it sounds like you are saying that sometimes not being able to recognize content is a good excuse, sometimes not

Jan: example captions – you should be able to turn captions on off. This is fine if you know what our captions, but if div and don't know it's captions, shouldn't be expected to
... upper-level condition on whole thing just like WCAG

Eric: and it ought to be handled in one place

<jeanne> +1 for one place to handle the not applicable.

Jan: yes

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.137 (CVS log)
$Date: 2013-02-07 19:38:31 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.137  of Date: 2012/09/20 20:19:01  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Found Scribe: KimPatch
Inferring ScribeNick: KimPatch

WARNING: Dash separator lines found.  If you intended them to mark
the start of a new topic, you need the -dashTopics option.
For example:
        <Philippe> ---
        <Philippe> Review of Action Items

Default Present: Jeanne, EricHansen, Jan, Greg_Lowney, Jim_Allan, Kim_Patch, sharper, kford
Present: Jeanne EricHansen Jan Greg_Lowney Jim_Allan Kim_Patch sharper kford
Found Date: 07 Feb 2013
Guessing minutes URL: http://www.w3.org/2013/02/07-ua-minutes.html
People with action items: 

[End of scribe.perl diagnostic output]