- From: Andi Snow-Weaver <andisnow@us.ibm.com>
- Date: Wed, 15 May 2002 12:31:10 -0500
- To: w3c-wai-gl@w3.org
Attendees: Gregg Venderheiden Jason White John Slatin Gian Sampson Wild Cynthia Shelley Ben Caldwell Matt May Eugenia Slaydon Bengt Farre Loretta Guarino Reid Andi Snow-Weaver jason - 2.1 status. proposal by Gregg on the mailing list. Some discussion on mailing list. gregg - CMN comments - everything should also be operable via a mouse or via keypad on the phone. Mouse users use onscreen keyboard which falls under keyboard navigation. CMN thinks all pages should be operable via the phone but most are already. gregg - operable via keyboard. issue is what kind of keyboard. even screen reader users us the numpad for screen reader navigation. text entry plus "stepping" and "go" function - is this too restrictive? should we require arrow keys. CMN thinks we shouldn't require arrow keys. andi - where do we draw the line between browser responsibility and content author's responsibility? gregg - depends on technology. jason - serious problems with anything that ties it to keyboard interaction. checkpoint needs to be abstract enough. jason proposed success criteria on mailing list. may have to work backwards to arrive at checkpoint wording. gregg - if too theoretical, web masters won't know what it means when they are actually implementing the page. jason - technologies, especially w3c technologies, are implementing "these" already. concerned that if wording isn't right, sends mixed messages to developers. proposal including "text entry", some kind of "navigation", and some kind of "activation" is close. proposal should include use of device independent methodologies provided by the technology being used. cynthia - does this cover the "event" piece or just the UI piece. i.e. "onfocus" ? jason - if "focus" event defined by the technology that is not device-specific, that is the one that should be used. gregg - what is definition of "device independent" cynthia - two floating around W3C. one deals with input devices, one deals with PC vs. phone. gregg - have to be able to define it in more than just HTML terms. jason seems to be getting close to it in 2nd of his two points. jason - 1st point talks about what makes an event device-specific in a way that we don't want. if parameterized based on physical state of input device, it is problematic. i.e. coordinates, specific keys, switch settings, etc. 2nd point - if logically abstract events don't exist, use device-specific functions in redundant way. gregg - what is it that we are trying to achieve or prevent? typing on keyboard or using an alternate method for typing on the keyboard such as speech input. concept is all about entering text. has to be mechanism for moving among choices and activate them. can't do direct manipulation interfaces because they require eye-hand coordination. also has to be completely operable with mouse because more efficient for some people. guideline needs to be more general, then checkpoints can address more and more devices. jason - should be possible but not necessary to step through the choices one at a time in order to activate them. cynthia - minimum is that you be able to get to everything on the page in a device independent way. next step is more direct way to get to important things on the page. is that what you are saying jason? jason - not necessarily. must be able to invoke any function, do text entry, and determine what the choices are. need to re-work the proposal. where technology doesn't allow this to be done in device independent way, have to provide redundant device specific methods. gregg had some interesting ideas for level 2 and level 3 success criteria. <loretta leaves> andi - what if technology doesn't support more than one device-specific input method cynthia - like a kiosk with only a touch screen jason - so have to rule out technologies that only support one type of input device. need to be careful to distinguish between programming technology the author has control over and the hardware the page might be displayed on. john - asking that authors do whatever is in their power to ensure that their content is rendered in as many ways as technology allows. cynthia - unclear about whether wcag is concerned with independent input devices or independent rendering devices jason - both but this checkpoint is about independent input devices gregg - think problem is that we are trying to be generic in this set of guidelines. hard to do that with this one. "step" and "choose" is the simplest way to figure out what the choices are and choose one. jason's idea good: "text entry" plus mechanism to identify all the things you can do plus a way to invoke them. but what if the way you do it is with a mouse? breaks. jason - if working with mouse, can't identify what the events are. no discrete list of them. response is different depending on coordinates of the mouse click. events are hidden - no way for AT to list them. gregg - what if have screen full of buttons but they can only be activated via a mouse. so if publish API on how to access but there is no AT that supports it. jason - if don't use "standard" way of doing something, fail another checkpoint. gregg - what if create something that can only be operated with a mouse but provide API to programmatically activate things. is this accessible even though no AT supports it? cynthia - AT always lags behind technology. vendor has done everything they can do. gregg - government must make sure workplace is accessible. if no way to make a certain technology accessible, government shouldn't purchase it. cynthia - to make something accessible, there are a number of things that have to be in place - user, author, browser, AT, OS all have responsibilities. are you saying that all software should be published in technologies that are currently supported by assistive technologies. gregg - government is going to buy new piece of SW that only works on XP and 2000. government can choose not to buy it if they require it to work on older system. customer decides. cynthia - seems like someone could create content that meets WCAG but someone uses it with browser that doesn't meet UAAG, then it wouldn't be accessible. gregg - but it would meet WCAG. cynthia - sounded like you were saying a single vendor owned the content and the user agent and that they had done everything to make content and UA meet guidelines but AT doesn't yet support it. gregg - let's go back a year or two. if an author did everything possible to make a TIFF file accessible has nothing to do with conformance. cynthia - scenario described included both content and rendering tool. took a while for screen reading software to catch up with browser. w3c was encouraging people to use newer technologies that had accessibility built in even though the screen readers didn't support them. gregg - if "want" to use IE and screen reader instead of HPR. HPR is reasonably priced so it is not unreasonable to require user to use HPR. can people choose any screen reader and any browser? no. is there a reasonable screen reader and browser combination that they can use? minimum should be that 1) it is accessible using a reasonable and existing combination of AT (not too expensive, don't have to compile yourself) 2) not something that is unreasonable to ask authors to do. gregg - level 1 is critical and reasonable, level 2 is important and requires more effort, level 3 is more important things that will be critical for some users, other category (advice) - don't know when it applies, just want to leave it in. jason - need someone to write proposal for 2.1 gregg - interacts with topic we just did. put hold on it until close to end of call. see if there is a volunteer. cynthia - like idea, wondering how it interacts with testable. gregg - testable is a third facet. jason - way we're doing it now - assert that you have reviewed it - can be strengthened. review with additional "advisory" points in mind. can prove that took them into account. gregg - can test assertion by looking at the page gregg - author asserts that they have reviewed the content bearing in mind the items listed in the "factors to consider" section and that they have done xxx or have taken those actions they felt were possible or appropriate. that is testable. if we say they must have a process behind it, that is not testable by looking at the page. gregg - don't want our standards to say that effort rather than result is good enough. only use this for things that we really can't develop objective criteria for. companies cannot assert things easily, in many cases not at all. an assertion like this at the minimum level is not acceptable. liability per implied warranty law. cynthia - minimum is what is likely to be incorporated into regulations. gregg - may go to level 2 in some categories. if don't give them a minimum on each one, may lose that item altogether cynthia - they won't include anything that is not testable (like 508) gregg - everything in level 1 has to be testable. want to hear from others about this scheme. gregg - conformance levels are determined by three factors 1) testability (at least levels 1 and 2, think it has to be all three by the rules) 2) importance - most important (both how important it is to the concept and how it applies across the range of people affected by the checkpoint, like to stay away from quantity affected) 3) practicality of doing it for all content on all pages on the web cynthia - does that include tradeoff situations such as clear and simple text and use stuff designed for accessibility but remain compatible with old technologies gregg - first example doesn't exist because requirement is "as is appropriate to the content". 2nd example - if have to create two different web sites, not practical. on clear and simple language, level 3 could be an alternate version like a summary or something. level 4 might be a "tuned" site. john - like idea. not clear on how it relates to discussion about the conformance statement with testability in it and about minimum level not guaranteeing "accessibility". would like to see proposal in writing. gregg - will follow up with one. cynthia - like general idea andi - like idea in general but concerned that without defining minimum requirements in terms of technology used, hard to get consensus on when something meets the guidelines. also concerned about discussion that minimum means it works with existing AT. several weeks ago, discussed case of accessible Java application but no AT supports it. agreed at that meeting that it would be accessible. bengt - like it gian - good idea ben - like it. opportunity to put in more ideas than we would be able to otherwise. emphasizes importance of checkpoints we are asking everyone to adopt. eugenia - like it gregg - jason and gregg will work on 2.1 rewording. ben will help. jason - agree with first two. problems with reasonable/practical idea - what is reasonable to do may vary greatly depending on the technology used. what is difficult to do today may be much easier in the future. levels should not be distinguished based on assumptions about the resourcefulness of the implementor or the technology the implementor has chosen to use. gregg - reasonableness doesn't go into decisions about the checkpoints, only about what goes into level 1, 2, or 3. jason - "reasonable" opens us to charge that there was compromise going on behind the scenes in the working group. gregg - agree that judgements need to be objective. over time it should be easier and easier to comply with more and more. Andi andisnow@us.ibm.com IBM Accessibility Center (512) 838-9903, http://www.ibm.com/able Internal Tie Line 678-9903, http://w3.austin.ibm.com/~snsinfo
Received on Wednesday, 15 May 2002 13:33:56 UTC