Minutes from 11 December 2000 ERT WG telecon



Update from F2F

LK I'm reconstructing notes from my palm pilot. MOral of the story: don't replace batteries on a crowded convention floor. The bottom line is: we expanded the document to cover evaluation in general. We're including an indication of where particular judgements come from. Including if the source of checkpoints, what it is. If a human, what judgement was made. We're also separating the specifics of checkpoints (or whatever they are called) were passed from the checks. Therefore, a log of the raw data. i.e. this checkpoint passed this one was not. In principal, we decided to covered things that were not immediately well-formed XML. But, we did not get to the details of how to do that. We made a first cut at a pseudo-RDF diagram of circles and lines of the data structure. Subsequently on the list, Al has posted something that formalizes without getting into the angle-brackets what an entry would be.

WL I don't have any corrections, but to supplement. Al's pseudo-code is pointed to from the agenda. The most significant move was typified by the change in the name. It represents a tentative expansion of scope as to reach into other areas other than Accessibility into all W3C areas. This will take cooperation between PF and technology architecture group.

LK Is there any W3C group that is looking at a higher level review.

WC Quality Assurance activity internal to W3C. Prevent inconsistent support, like CSS.

WL This project could be a tool for this person/group?

WC Yes. I think this is what Daniel was pushing for.

LK In case of CSS, requires a human to look at? Say rendering of italic. Have a reference implementation, have a tool that would do pixel by pixel comparison with proposed implementation.

SL I believe the software conformance for VRML are done that way. The conformance test suite is done visually.

WL In order for them to claim CSS, they would have to qualify.

LK It looks like they would have a need for a human component even in principle no human judgement is called.

WC Right, and that's one component of EDL.

LK Right, so adds support to the human component of EDL.

WC Minutes should be out this afternoon.

LK Speaking of PF, is there anything can say about those meetings?

WC Primary topics: SMIL, XMLGL, and TAG.

LK What are going into the XML guidelines.

WC I talked at XML 2000, I'll point the group to my slides. /* quick overview */

LK Ultimately, EDL besides describing a finished product could also apply to a description of a language. If you had an XML language defined by an XML Schema, point into the schema. E.g., where the schema does not show how to provide text for a media object.

WC not in the initial scope, but in the ultimate scope. An issue with the XML GL is final form languages like SVGFO. No separation from presentation and structure.

LK How would you say in EDL that the presentation is not separate from language.

WC Not my question, but PF issue as to how to handle.

WL GL position will be to separate church and state.

LK That discussion is going on in GL (what is presentation? what is content?). I gave out a simple example, results were 50/50 (if it were presentation or content). Will that have any bearing on what we're doing?

WC Don't think we need to wait for that decision.

LK Al's proposal.

WC those of you who did not attend, any questions at this point?

MC What is the scope? Is it intended to be fine grained and say "this element on this page conforms with this and this but not this and this." or in general, "this and this ok this is not" or most general, "this page conforms to Level A according to this eval tool." I'm not sure level this language speaks at?

WL The first.

LK Right. Also, there could be times when a checkpoint does not just apply to one element or region but to 2 or more. e.g. an image map. There is a p3 checkpoint in wcag 1 that asks you to use redundant text links. conceivably 2 pointers that are not right next to each other.

MC Is the language intended to be for automatic checks or that the user has responded?

WL All of the above.

MC How is the presence of meta data on a page expected to be used beyond the original tool. There is a lot of detail i'm not sure what to do with.

WL When we see, get involved with, and meant for machines, it boggles my mind. A lot of this depends on future implementations that mine the RDF data.

LK Here are a couple examples

  1. A tool like the WAVE that displays the page w/flags showing where there are problems. You could have the output used by the tool to add additional notations. an evaluation tool.
  2. An editor could use this to pick out problems in the code could show mistakes in line - a flag in the code.

MC An out of line evaluation then back to authoring tool for fix.

LK Yes, the challenge will be to have this ready in a time frame for authoring tools to implement. I talked with Macromedia a the meeting. How much work to bring in the outside stream given my deadlines?

MC Yes, and a call for using RDF. A tool like Macromedia parses XML. If tools already read, then only need a lightweight module to read RDF/XML. My concern is that will we put energy into something that won't be used.

WL Hotdog already implements it. If you have it in accessibility mode, it will put in markers.

LK Using internal checks?

WL RIght.

LK Dreamweaver also does internal checks.

WL And HoTMetaL.

MC Forth possible implementation, something recognized by search engines, filters, etc. that you can say, "only give me pages that have pages w/alt text." Good use of meta data.

LK Will EDL have use in proxy/repair facilities.

WL Conclusion from the meeting is that as long as its optional/configurable there is no reason not to work on repair side. i.e. don't impose repair but you assist for someone that wants to.

WC Other questions?

CR Makes sense. Still processing. Concern is how to push to the rest of the world? Will it give accessibility a bad name. If you put RDF Into your doc is the first thing accessibility info.

WC It's an open issue as to where the data is stored. Don't assume its in the source resource, not yet anyway.

LK If navigation is not consistent between pages, there is no clear page to put that on.

DB No questions now, but I'll look over the minutes.

Al's proposal

unit evaluation record - Al Gilman, 4 Dec 2000

WC He's saying, however you define unit, here's how you can track an evaluation for it?

WL Yep.

WC applicationOf seems to refer to URI if on the web, or File:Save if authoring tool function.

LK applicationOf and whoSez apply to each test?

WL No, those are likely default at top of doc, but as go further there may be specific ones that aren't.

WC Let's talk with Al on the list, and get him here next week?

WL It's a template. It will serve as a minimum for what you have to do.

LK We'll pursue that will Al.

HB It suggests that there are only small portions of any doc that are to be evaluated. We can rule out general paragraphs. This is only dealing with active elements.

WC Don't think you can say that. WCAG deals with all content.

WL Can set up, not dealing with all content.

HB Hate to see it deal with simplicity of language.

WC But this is to track human judgements as well.

LK There is the XML serializing of RDF. Also simpler notation in curly brackets: subject, verb, object. I suggest that we continue on the list to come up with those then see where we are.

WC Lots of open issues, e.g. from Len's email/Sean's summary: where does this go (Rec? Note?), where store this info?

WL Until something there, can't be sure.

LK Authoring tools is a potential consumer of this. The usual time to meet with AU was last week. When's next time?

WL How about first Tuesday in January. We have a meeting tomorrow.

LK Do they have people lined up to implement ATAG?

WL Yes.

DB Lisa would like hear more about it.

WL Example of storing info is at UWIMP. It's a questionnaire, it generates a header for your document.


WL Yes and no. HTML/XHTML can't validate with RDF, so it points to a statement from the head.

LK A subcommittee of folks from F2F formed to do actual implementation of RDF. Therefore, talk about it on list.

WC On list for archive purposes.

LK Let's get started on that. It was me, WC, Charles, Daniel, Sean.

Action subcommittee (LK, WC, CMN, DD, SP): first set of RDF triples or schema by next Monday.

WL We do not want to lose sight that XHTML is XML.

WC right, that's why it's in the initial scope. Using Tidy to transform HTML to XHTML. Likewise, talked about how to canonicalize ECMAScript so that can deal with.

Action LK: straw proposal for how to handle languages not in XML.

$Date: 2000/12/11 17:47:02 $ Wendy Chisholm