11 September 2000 ER WG Telecon

Summary of action items and resolutions




Agenda posted 8 Sept 2000.

HTTP Refresh

WL Never got an answer about this.

LK If constructing HTTP header from cgi script, could put meta tags in HTML. Just using HTTP equivalent in HTML.

WL My problem is that this is an area that WAI does not seem to address there is anything at this level.

DD I think it addresses refresh.

WL But how do you handle it when it is in HTTP rather than HTML.

DD As the author you handle it the way you want. It should not be there.

LK How does a tool like WAVE handle it? You can send a request to get the header itself.

WC Does something need to be added to AERT?

LK It's written for HTML tags, this is not an HTML tag. You need to look at the HTTP header.

Action LK: Write proposal for AERT to check HTTP header for refresh issues.

BM We're well aware of redirect from headers. As long as servers do it, no worries.

LK But if redirect after time-out...

WL Or if it is recursive.

BM Yes, there are pages that you get stuck in.

LK Pages could flash causing flicker.

BM Yes, and not friendly to screen reader.

WL Could be a slide show, but URIs in there. Put up blank HTML and send screens. Could probably do a movie once broadband arrives.

LK Could be a way to push content into a frame, e.g. stock market ticker. Refresh every 10 minutes.

SVG and RDF tool

DD Release of tool programmed by student over the summer. The idea was to take an SVG file with RDF metadata and produce a textual version of the image. First developed vocabulary to describe generic graphic: connections, positioning, etc. Then a reader of SVG that would parse information expressed in RDF. In addition, the student worked on an editing/help tool to produce the metadata. Not an SVG editor, but an add-on to help the author to augment graphic file. Using menus, it will let you choose which vocabulary you want to use.

LK Given a typical SVG picture, how many semantics are in there already? e.g. if you have 2 circles and a line between them is there an SVG way to say, "these circles are connected by this line?"

DD SVG does not have a notion of connection. Grouping, yes. It is in RDF and therefore added to SVG via the metadata statement. We use RDF and property of our vocabulary.

LK If you have 2 circles, with a line between them, and the semantic that the connection exists. This become part of SVG file?

DD Use the LINK structural property in SVG. Create id's for those three elements and express in RDF description.

LK The RDF is in a separate file?

DD It can be internal or external. Metadata can sit anywhere. SVG has a metadata element so it can sit in the file.

LK At this point, say you have 20 circles and there is a line one end point at one circle, other end point at another circle. Imagine a heuristic that could figure out that it was only affecting those 2 circles not the others. Is this part of the tool?

DD Look at the vocabulary to determine if it is possible to express that relationship.

LK I'm not thinking of the completeness, I'm thinking that circle 11 is connected to circle 12. The tool could suggest that the line connects 11 to 12 and not say anything about the other circles.

DD The tool does not have any heuristics like that. It imposes constraint on the wording you use and then tries to produce correct sentence.

LK Is it French or English?

DD English. I haven't played with it too much.

WL Worth sending to RDF IG?

DD I did. I sent it to several groups.

LK One of the things we've been talking about is using RDF to express accessibility or lack thereof in HTML. One could imagine taking a similar approach.

DD SVG used to have a metadata attribute.

LK If we wanted to use something like this to make accessibility statements about an HTML file is there something we would want to add to XHTML?

WC Could do something similar by using meta tag in head with id elements.

WL But we want to make claims about files without modifying them.

WC So we use XPath, but then information will become invalid when file changes.

LK missed...

WC If repair tool, like A-prompt, could add metadata right to the file. Otherwise, could take snapshot and add metadata to that file. However, does not solve problem of updates on that page.

LK Want to get back to SVG.

DD The logical ideal step is to formalize this RDF vocabulary so it can be integrated into tools.

LK You could have specialized vocabularies in different fields.

DD In PDF we've been working on classification of graphics on the web. Numerical, conceptual, etc. Numerical is made accessible by providing the data rather than a description. Conceptual does not have formal data behind it. This tool can help formalize the relationship.

WL like data table vs. format table.

DD right. Scope and heading is similar to SVG stuff.

LK To what extent can we pursue this in ER? If we think of it as a forward-looking repair tool...

WC What feedback do you want? Hammer at the vocabulary?

DD Yes. Think about graphics and if it is adequate. Charles has already commented on it saying that it is complex and we could create a simpler core.


DD LK you are blocking out the week. Do you have more detail?

LK I'm going to talk on the 5th and travel to Pennsylvania evening of 4th. Then, could come back Tuesday night if necessary. What's blocked out is Monday night and Tuesday.

DD If we had ER on 4th then PF on 5th. The issue is that we have AC meeting ending on Thursday. I was hoping to piggyback on that. The weekend is not good for some people. I don't want to wait until end of XML conference. If we can do something at beginning of XML week that would be good.

LK I'd like to be around for PF, but something has to give.

DD WC you may be presenting at XML, not sure yet.

WL So our face2face is when?

LK Undetermined. Washington first week of December but we're not sure when.

WL At the moment looks like the 4th is ER, and the 5th is PF.

LK Let's continue this over e-mail.


LK Is the mission statement the goal or the activity? It currently has a couple of activities. This describes output. JB suggests that we just discuss what we're doing not the goals or success criteria.

WC Could move to success criteria section (section 5).

LK Could just remove those, already in success criteria.

WC JB suggests:

developing techniques that describe how to evaluation and retrofit Web pages according to WCAG 1.0; reviewing evaluation, retrofitting, and transformation tools; tracking tools and related resources and maintaining reference links to these; developing a limited number of tools where no appropriate tools exist.

sounds good to me.

LK Some of that is in scope and deliverables.

WL Our mission is to make the world safer for accessibility.

DB Should be more of a summary.

LK e.g., to facilitate evaluation and repair of web sites.

Action WC: Take suggested mission statement back to Judy.

WC Facilitating in 2 ways: writing own tools and documenting techniques for other developers.

LK Coding, listing, etc. Does anyone feel that mission statements should not be that abstract and simple? Should they be more specific?

DB Fine with abstract, details elsewhere.

HB Keep it simple. Detailed constrain us in unexpected ways.

Unanimity: Keep the mission statement simple.

LK Success criteria. Current are impact related.

WC don't agree that it is "indirect." There are a number of tools being developed by members of this working group. AERT affects tools indirectly.

LK We have "increasing the number of tools" perhaps should say, "Increasing the features or quality" of tools.

WL Or at least affecting.

LK I agree that it is harder to measure success in terms of impact. If haven't had impact, what else is there.

WC The other bit is that we produce the deliverables that we say we will.

Action WC: suggest following success criteria to JB: increase number of tools for repair/evaluation, increase the quality/features of existing tools, deliver the promised deliverables.

LK She also comments on deliverables. We're saying we may do this but we are not committing that the group per se is coding.

WL The question is whether it is our contribution or members of the group. What's the difference?

LK Right now, the folks at CAST do their thing, Chris does his, I do mine. Whereas the AERT is a product of all of us. We all contribute ideas and words. There is not someone out there developing AERT and we review it now and then.

WL Hence "related."

LK Thus, to be analogous, we would have a meeting "how are we going to architect this tool?" and related questions.

WL Do you want to specify that that won't happened? It won't happen.

LK Hasn't happened yet. We don't want to rule it out.

WL We're not designed for that.

WC If we had students or if we all interacted with schematron or some RDF database in some way, then we would have to interface. Perhaps Bobby, A-prompt, and WAVE will all store metadata the same way through a piece that i might write. So, don't see us working on one tool, but perhaps pieces that we may all refer to.

LK Can we wordsmith this? Schema as coding? Coding doesn't have to be self-contained module w/user interface?

Action WC: wordsmith deliverable section of charter to make clear that coding could include development of schemas.


LK Any comments?

WC Have not played w/latest release.

DB Me neither.

HB Me neither.

WL It has the opposite approach to Bobby in that it marks what is there rather than not there. That is valuable since it is like positive reinforcement, but it may be hard for some people to understand. You might want to explain that difference in philosophy in the paragraph before where you input the URI.

LK Have there been things that have been particularly helpful? or things you have missed?

WL Thinking of it as a validator may not be the best way to think of it, but that's how I think of it. As a validation tool it has not uncovered anything in the pages I have written.

LK It gives you the info for you to uncover problems.

WL That is important for people new to this, but it needs to clearly state its intent.

LK Many people ask, "what does it take to be WAVE approved?" or "what icon can i use?" Doesn't make sense for this.

WL Similar to "this is the tool that i used." could be a big plus.

LK 2 things that WAVE does not handle right now: 1. analyze javascript in more detail - is this benign javascript or cause an accessibility issue 2. doesn't do any repair.

WL Javascript is more groundbreaking.

WC Javascript for sure.

HB How many other scripting languages are there on the Web?

LK Flash.

HB Is that analyzable in a similar matter?

LK In principle.

WL Believe they provided a tool to help with that.

Action LK: Find out if Flash is analyzable or what tools are available to analyze Flash.

HB What about DHTML?

WC That's just CSS, JavaScript, and HTML.

HB Not Microsoft proprietary?

WC Nope. See that this group more equipped to analyze new JavaScript for functions and patterns and feed back to WCAG for techniques doc. We know that <a href="javascript:"> is bad, but many other patterns, functions, and common uses that need to be analyzed.

WL Want to feed back to WCAG "name" being missed on A. The implication that name can provide a target for accessibility.

WC Putting name on everything so that can link to anything?

WL tabindex and accesskey are being shorted. There hasn't been enough explanation of techniques for using.

WC Results from test pages don't show how to clearly use.

LK What is tabindex useful for?

DB Creating a order that matches the visual order.

LK That should come for free.

DB If you have something that 3 columns wide, which way should go alphabetically? Might be better to go across or by column. Just have to make it consistent.

WL You can chunk things, skip over blocks, then go back to them.

LK For someone who is blind, they have a concept of the order of things, then give them a tabindex that's different that would be inconsistent.

DB How so?

LK Say you have things in columns. The reading order is by column, but the tab order is across columns.

DB That depends on how people use screen readers. Many people will tab rather than read.

LK But, say someone does listen by tabbing to hear one order, then they listen straight through it won't match.

WL In addition, the scheme we are discussing, but how to do it is something else that needs to be discussed.

Action WL: write something about accesskey and tabindex.

WC We got rid of most of the discussion of tabindex in the HTML techniques b/c it did not work as expected.

HB I got feedback from XHTML write-up saying it was a mistake to get rid of accesskey and tabindex.

Action WC: send URIs for accesskey and tabindex test pages, as well as URIs for threads that discussed that.

$Date: 2000/11/08 08:17:23 $ Wendy Chisholm