WAI Authoring Tool Guidelines Working Group
Chair: Jutta Treviranus
Date: Tuesday 19 September 2000
Time: 2:30pm - 4:00pm Boston time (1830Z - 2000Z)
Phone number: Tobin Bridge, +1 (617) 252 7000
The Latest Draft is the Recommendation dated 3 February, available at http://www.w3.org/TR/2000/REC-ATAG10-20000203. The latest techniques draft is dated 4 May March, available at http://www.w3.org/WAI/AU/WD-ATAG10-TECHS-20000504. The latest draft of the Accessibility Evaluation and Repair Techniques is dated 22 August at http://www.w3.org/WAI/ER/WD-AERT
JT When are we planning to look at new guidelines. Judy asked if we wanted to start work on new guidelines at a particular time.
WL Other groups have a time, but we have a better document.
JR We could put out a 1.1
CMN Do we want to make it explicit that we might make a new version?
JT Do we want to make it start earlier
CMN We can start on a wroking draft of requirements tomorrow if we like
WL If WCAG changes we will need to change
CMN but only the reference to WCAG 1.0
WL And the techniques
CMN True. As I understand it we won't need to change the techniques themselves, just the organisation. Propose to put a possible revision of ATAG into scope.
JR we could put in new milestone
CMN date needs to be variable on WCAG, at least
JT That can be done
Resolved:
add them
JT have people seen the new techniques draft and evaluation techniques draft? They are the areas we are going to be working on during the face to face.
CMN Control logic, or flow of interaction - not relevant to plain documents, but relevant to site development, interactive and interactive multimedia content.
JT would include navigation structure, templates for sites
CMN and applets, interactive pieces
JT applets and so on might be seperate from site nav controls.
CMN We could split it a number of different ways, but it is currently under-represented.
JT where would accessible templates go in there?
CMN I don't know if it goes into that straight away. I think the answer is to use AERT for templates. (Except I think it only works for HTML)
JR I think handing off a big test like templates to AERT requires that we have an AERT tool. It would be hard if it had to be done by hand
CMN But in the absence of a tool we should still point to AERT
JR Sure, but it is nicer to have a tool than to have to use a spec
CMN Sure. But the spec is better than nothing
JT It would be good to have additional wording
JR We should have a process that takes less than a week
CMN If we don't do a complete test then there is no value to a conformance claim
JT Random testing is well established
JR It is possible that there might be a bug that causes a particular anamolous result.
CMN There is also a well-established tradition of mathematical proof, and that is used in the software industry
JR We want some kind of statistical reliability rather than a proof. If it is the programmers we can expect them to know better what their programs do but for someone outside it is harder. For example, "if an image is known with certainty does the system make suggestions from previously written alt text" and gives a toolbar generator. There are a whole bunch of things that the tool developer will know. Do we mean everything that is done?
CMN Sure. If there are inaccessible parts then the site isn't accessible
discussion of randomised statistical testing vs mathematical proof
JT We should discuss this further - do we want to do randomised sampling?
JR How randomised?
JT Probably we will give sample tasks for the tool - we are doing random selection of what the test will entail.
CMN The process we use is proof by contradiction - unless the developer makes some special hack to meet our specific requirements, it is a reasonable method, and the odds of that become vanishingly small
JR With a reporting tool anyone can add an assertion that a tool fails a particular requirement, which gives a larger randomisation.
JT So we should talk about this at the face to face. Any other gaps?
CMN Another part is the way that the two documents relate to each other.
JT Would it be worthwhile to link the techniques document to the evaluation document at a finer level (per-technique or per-guideline)?
JR We could have pointers for more ideas
CMN It might be easier to point from the evaluation doc to techniques
JT Techniques should have conformance info?
CMN As Kynn used to ask, we should say whether a technique conforms, or is extra, or what
WL A tool should have a way for crediting things that are done in addition to meeting the requirements (for example having a user-configurable prompting schedule).
CMN It is not tested in our current scheme, but it is useful
JT What ranking scheme we are going to have is something we would like to talk about - stuff for consumers to use the information.
CMN So maybe we should ahve techniques that are "extras" and check for those as further information.
JT Once we create a reports generator we may want to give consumers a way to search based on customised requirements.
WL We will be discussing this in Bristol, right
JT Yes.
CMN The most obvious extra is the operating system for which the tool is available.
WL There are probably a bunch of other things that are relevant.
JT Please read the new techniques document and identify where gaps are.
JT If people are attending the two face to face meetings they could consider coming to the Device Independence Authoring Workshop immediately before.
WL I imagine anyone coming would be interested in that.
JT Right. Part of the registration is a position paper - so don't get surprised.
CMN Is anyone from Microsoft planning to come.
LG I can't, but I'll follow up later.
CMN It is very immature - unless people have read the discussion and the document it is too early to introduce. I might be ready to talk about it at the face to face.
JT Where do we go with this.
CMN I guess it will end up as a basis for specifying a testing tool. The idea is that it will combine test results, and provide a testing report tool.
JT Can you explain how it relates to the testing stuff that Jan has written up?
CMN Not yet. I am trying to work out how to relate the two.
JR I am looking at describing some tests that can be done as a yes/no test.
WL That is the hard part
JR I guess there is a certain part of "I'm not sure".
CMN I think we are all approaching from different parts of the puzzle. I think Jan is looking at what tests need to be done, and that mine will get that as a filter and will apply it to a set of tests...
WL I have tried to put together a bit of a matrix of what tests to apply to what tools
JR Are there any other areas that we need to look at in addition to those that William has?
JR I guess frames
WL Forms seem to get a lot of attention.
CMN A more consistent logic within this document would be good.
Action all: Look at both documents and look for gaps, and for how to make a consistent logic within the document.
JR By the way, when it comes to image editors, how are we suggesting that they include equivalent alternatives? In some types of image you can put them in the text track. It might be better to also have them saved seperately. How are we saying they should do that.
CMN Good question. (But I don't have an anser, although I think the RDF-photo note is a good start at an answer)
Last Modified $Date: 2000/11/08 08:13:13 $