HB I've been playing with Lift.
LK Concerned about some results it gave me for a page.
WC Paolo and Gorgio were at the f2f in Amsterdam. Said it is very customizable.
WL It needs templates, one for accessibility, one for usability, etc. based on what you're doing.
LK A fixed way of expressing information. The syntax is locked down but people can come up with different terms.
CMN A way of making statements about objects. Uses URIs. You can feed a pile of RDF statements to a tool and it will go collect the information from around the Web. All it does is let you say, "X is Y to Z." "Charles is brother to Will." Neat way to build databases. Once you define objects and relationships, put them on the web with a URI. Others can use those objects. "See also" is simple reserved relationship that exists in RDF.
LK "asserted by" is another.
CMN Neat way to build distributed databases. There is already a scheme for checking conformance to URIs. Level A, AA, AAA, N/A. Trivial to apply to any of the guidelines or checkpoints that have a URI for each. These are the rules for deciding conformance. Feed it WCAG or AERT checkpoints instead of ATAG.
WL Also has meaning in PICS.
LK Applying it to WCAG, they have anchors in front of each checkpoint but they don't have a SPAN or DIV that wraps around the whole checkpoint.
CMN If you want to do database processing, that gets tricky. As soon as the guidelines are in XML or XHTML that becomes less of a problem.
LK As far as the AERT document is concerned, it is made of ...
WC Using to store state info. Chris and Michael's ideas should be discussed.
BM We should be able to store things once.
LK Stored locally with the page as part of the authoring tool or publicly available on the web or both?
BM Ideally carried with the page. There may be cases where not possible.
CMN Ideally do either. e.g., developing a site, keep it private. Once published, may keep public.
LK In developer space, may want info about each element, but once published may only want per priority, as Michael suggests.
CMN When I was working on conformance schemes, I needed to know details because of the relative priority of ATAG. But, I also wanted to store, "Tool X is AA." When you get into partial tests and partial evaluations, I could get so far along and then someone else could come along with to finish it. You can then attach who evaluated which checkpoints and then determine how much you trust those assertions. Lots of tools. See the RDF home page.
WC Lots of talk on RDF, Chris has ideas about XML.
CR Save the state in an XML file. Not quite sure how RDF works into that.
CMN It's a data format in XML. I would urge you to use RDF for the format.
CR For example, we do an evaluation, "Does this require longdesc" they say "no" we could save that in RDF?
CMN yep.
LK If do this tag by tag, for this IMG there is a missing "alt." To what extent does RDF have built-in statements that will do that or will we have to define additional assertions?
CMN Define what you want.
LK Do we define an assertion for every checkpoint? Or "element X satisfies element Y" "not yet tested" "does not satisfy."
CMN Dig through the archives to find the scheme I posted a while ago. It builds triples.
LK tag relationship checkpoint. There are accessibility checks which are not a function of one tag, but a relation.
CMN Build a relationship, "depends on." Charles depends on Len, Len depends on Wendy, Wendy depends on Michael, Michael depends on Charles. It's circular. That cycle does or does not meet the requirement.
LK Can we work in terms a specific checkpoint that involves more than one tag? Inside a link you need either an image with non-null alt-text or a text link.
/* wendy goes to handle person at the door */
LK if accessibility has to do with a bunch of elements without a container, you define an accessible set of elements ..
CMN RDF has a "for each" and a "bag" concept.
LK You can define which roles various tags perform. There is a classic thing in AI 101 that says, "Socrates if man. Man is mortal." then ask if Socrates is mortal says yes.
CMN There are number of ways to do that basic logic. RDF only supports low-level logic
LK Is it equivalent in power to Prolog?
CMN It's a data format. It doesn't include the processing.
LK How should we proceed with this? Create a set of definitions?
CR There is a schema for WCAG?
CMN I wrote it for ATAG.
WC Found out about a new technique using classes and namespaces... main concern, Len?
LK If we have a general solution to use "class" then why use MAP as a hack? Why not push for the general case? In terms of simplicity, why bother?
WC What about multiple navigation bars on a page?
LK Take Yahoo, main navigation bar and another one. If use class no longer need map.
CMN The idea that you know what a class is, depends on having an agreement that it means something. There is an agreement that MAP means something. It's in the HTML spec. To say, "we want a class for a collection of links, why bother?"
WC Dan, Charles, and Ian conversations.
/* scribe note: Charles and Dan's exchange about classes and Dan's notes, "Grounding link relationships and classes names in the Web."*/
CMN I want to use URIs for classes so I can use RDF to describe the class. Still using MAP as basic element. If I then want to class it further I can do that. "class" is XML by stealth since you want to use HTML but you're inventing extra elements.
LK In other words, class is a kludge to bring in the advantages of XML w/out XML. So, what's the best thing to do in this kludge state?
CMN Best way to proceed is to have a way to say what classes are. RDF is what I would be using and ideally I would use a URI as the class name and be done with it.
LK Here's another kludge, for each class you could create a new attribute and that would would be mutable strings rather than whole RDF file.
CMN question of syntax.
LK If use class, still use MAP?
CMN Yes.
LK there are pages like Yahoo where they have main categories, and underneath "art, culture, etc." there are subcategories. The top level categories are useful. Wrap those in
WC The links on Yahoo are the content. MAP is for navigation bars at the top that may get redundant or in way of content. Not all groups of links have to be contained in MAP.
LK MAP is not defined specifically for navigation bars. Each link has shapes.
WC Are image maps usually in the middle of the content?
LK Old AT&T page had an image map on top, one in the middle to feature services.
CMN MAP is already in the spec, class is not.
LK I'm looking at the spec now, it says, "client-side image maps."
WC It says, "The MAP element specifies a client-side image map (or other navigation mechanism) that may be associated with another elements (IMG, OBJECT, or INPUT). " (from The HTML 4.01 spec )
LK Perhaps poor narrative, but it makes me feel that philosophically we are beyond the spec. Is this an oversight. E.g., CMN and WC say that the spec was modified, is it that they should have gotten rid of the statement that you must include a shape? Understanding the intent and assuming that what i'm reading is an editorial oversight, I can be more comfortable.
Action LK: send editorial comments re: HTML4.01 to the HTML working group.
HB If you are going to increase the font size of things in a fixed area, it will be more difficult to use.
CMN It doesn't provide that facility.
HB I thought LK was going to write something about the size of the MAP.
LK No, that you must require a size in the MAP.
CMN It has nothing to do with the rendering of the MAP content.
HB ok.
Resolved: Tomorrow with AU at 2:30 ET.
Resolved: At next week's meeting, scripting is the first item to discuss.
HB want discussion on merged glossary before send to other groups.
Action everyone: read HB's merged glossary and comment to the list.
$Date: 2000/11/08 08:17:43 $ Wendy Chisholm