W3C home > Mailing lists > Public > w3c-wai-ua@w3.org > April to June 2000

minutes from WCAG and UA joint telecon on 4 May 2000

From: Wendy A Chisholm <wendy@w3.org>
Date: Fri, 05 May 2000 08:01:07 -0400
Message-Id: <>
To: w3c-wai-gl@w3.org, w3c-wai-ua@w3.org
The minutes are available at: http://www.w3.org/WAI/GL/meetings/20000504.html

I have also included them below.  Note that as I cleaned up the minutes I 
found several points that needed follow-up.  Therefore, I assigned some 
action items <smirk>.  These are obviously negotiable since people did not 
accept them during the call. The only person who does not get off the hook 
is Harvey, as he accepted the action to take a question to EO.   A few have 
been "assigned" to WCAG WG in general and if everyone is supposed to do it, 
no one will.  Therefore, it would be appreciated if people volunteered to 
take them.

It was a very interesting call, but I'm not sure that the UA folks got all 
of the answers they needed.

UA WG - what further do you need from WCAG WG?


04 May 2000 WCAG telecon
Joint telecon with the UA WG to discuss Structural Navigation issues
Action items and resolutions
HB suggest to the EO group that they need to raise consciousness with 
advertisement aggregation companies (that's one place where alt-text might 
be getting lost).
WCAG WG investigate the proper use of DIV.
GJR show what a possible blueprint migration from HTML to XHTML to XML 
might look like.
WCAG WG create techniques that show how to use XML with a style sheet.
?? Is it the case that when style sheets are used to convey presentation 
that proper markup is used? Do pages that use style sheets use more 
structure markup than those that do not use style sheets?
WC find examples of ISO-HTML pages that are conformant (i.e., have properly 
nested headers).
WC check out ISO-HTML algorithm to determine structural divisions. take to 
WCAG WG review AG's amazon rewrite.
GJR re-ping glen gordon and mark hakkinen to find out what they are doing 
to skip over repeated text. could be simple comparison. good for techniques 
Jason White
Ian Jacobs
David Poehlman
William Loughborough
Harvey Bingham
Wendy Chisholm
Andi Snow-Weaver
Jon Gunderson
Gregory Rosmaita
Tim Noonan
Mark Novak
Al Gilman
Cynthia Shelley
Rich Schwerdtfeger
David Tanner
Gregg Vanderheiden
Dick Brown
Structural Navigation
JW Not sure who's chairing this one, if JG wants to our I should. Since 
it's during WCAG I'll go ahead. Congratulations to JG for receiving an award.
JG "Academic professional excellence award." nominated by the college. 
recognition by peers through the university. did not mention accessibility.
GJR so no minority opinions then?
JG they didn't talk to any UA WG folks.
JW essentially there is a longstanding issue with structured documents. if 
the document structure has been marked up correctly, one could navigate 
through it. knowing where one stands and then locate particular points and 
then browse the document by kinds of content. e.g. moving through an 
outline to read headings in order to provide an overview of the contents of 
a document. there are issues that relate to both UA WG and WCAG WG. What 
markup conventions should be used for the UAs to interpret? Two fold issue:
1. what can be retrieved from DOM. connection to hierarchy and XML tree 
2. use of appropriate markup so that the structure is explicit.
Al has suggested that the document hierarchy is not isomorphic to the 
navigational hierarchy, at least not in all cases. We need to consider UA 
capabilities and future versions of WCAG to make sure they are in synch. In 
SVG one has the advantage of identifying portions of an image. Concept of 
structural hierarchy and navigating it. Can apply to documents, images, and 
multimedia. Jon - want to elaborate on any aspects of the issues?
JG From what I see, there is agreement that the MAP element should be used 
for collection of links. My understanding is that "title" attribute of MAP 
should be a label for what the group of links is. The navigation issues are 
still cloudy. For example, I've got different stuff on the page. I use 
speech output. I go to Yahoo - navigation bars, ads, info, that is static 
and a dynamic area. When linearized, the static stuff is at the beginning. 
The 2 questions I have: can we mark up the 2 areas? MAP is clear, but what 
about a list of phone numbers? A New York Times article? When someone comes 
to the site they have to go through all of the links to get to the new 
CS would changing the page title help with that?
JG possibility. more specifically, is there an easy way for people to get 
to the new info?
GJR individual pages should have individual titles. using a hotkey, someone 
could quickly get the title. we should have invited henter joyce (HJ) as 
they have support for this. Jaws will skip over what it has run into before.
DP on a shopping site, and click "buy" it leaves you were you were so you 
can continue on. it's very helpful.
GJR another thing, that both Productivity Works (PW) and HJ do, is 
structured navigation. the ability to move by element or header. these are 
implementation scenarios. not sure if heuristics have to do with markup or 
other techniques. we ought to find out what they are doing.
JG that's my main question: is this a UA function? rather than require more 
work for authors, is there sufficient info for UAs?
JW depends on if there is structure in the markup. if lists, tables, etc. 
are not marked up then it will be impossible for the UA to compensate. HTML 
does not encourage the use of proper container elements to designate the 
scope of content. Even tho headings may be used correctly, the section to 
which it applies may not be clear since there are no specific container 
CS might also work to have something like tabindex for named anchors.
GJR tabindex is allowable to put in named anchors but it is not supported. 
HTML spec says that A element can take accesskey.
CS i want to define where people can go. it makes sense to skip over nav if 
not using it. i want them to go to ads.
GJR offering blind user same functionality as sighted user. how many people 
actually read the ads?
CS if you make it easy to skip over, developers will have issues with 
funding. but my issue goes beyond ads.
GJR two-fold process. the functionality could be used to skip "foo" it 
could be used to return to "foo."
JP the UA could use that info in the user interface.
GJR when we had the discussion on GL re: reformatting pages with Scott 
Luebking I said that would not fly. authors don't want ads on the bottom.
RS and they won't want you to skip the rest.
GJR i don't see the difference between someone who goes to Yahoo and their 
brain is trained to start at a certain place. despite things beeping and 
flashing, they ignore. Most people when they get to a page with a lot of 
ads will ignore.
DP I browse as a disabled user, one of my favorite things, is to go to the 
first field on a form. i don't even listen to anything up to that point.
CS my concern is not so much for other users of auditory interfaces, like 
the car computer, people who are supplying content will want their ads to 
be read. the user and the author should have some control.
GJR yes, we need to be careful. there are times when you want that info.
JW no user should be forced to read anything on a page. as a print user or 
speech user they should have a choice. it should be possible to avoid or 
return to particular pieces of the content and to move through the 
structure as quickly and easily as possible.
GJR I am loathe to use comparisons to hard-copy print media, but when 
people read magazines, there are lots of gutters on the far right column. 
some people go specifically for ads, some people don't. this is obviously 
an issue wrestled with in other media. We want to satisfy both sides of the 
equation: skip over and return to.
CS that's reasonable. my initial point was that i'm not certain that 
existing markup handles this scenario rather than trying to find a kludge.
GJR interested if the kludges that are already in effect could point us in 
a direction to solve the problem.
JW where is the markup missing? DIV could be used to denote block level 
content. title could be added to it for human readable characterization.
CS not saying that there are missing feature, but that we ought to explore 
it. DIV makes sense. but is tabbing between DIVs supported?
GJR but we can consider it here. Al has floated such a proposal. Here is 
navigation section. here is content. etc. Ads are routinely skipped over by 
people who are blind because there usually is not any info. You usually get 
"click here" or file size. By not assigning alt-text to areas the areas 
become useless.
JW we have techniques to markup image maps and correct markup for groups of 
links. i don't think we have any discussion for the proper use of DIV 
elements. links, tables, forms, are there because of distinctive markup and 
can be readily identified. At one level, provide appropriate container 
elements (DIV in HTML, group in SVG) there is the larger issue of whether 
there are cases that even if this is done the tree that occurs in the DOM 
corresponds to a good navigation tree. is there a need to provide support 
for navigational features. that has an impact on UA capabilities more so 
than markup. question of relationships that are not reflected in document 
tree that need to be represented. AG raised it in his message.
@@WCAG WG investigate the proper use of DIV.
AG want to respond to CS question. i don't see any difficulty in devising a 
markup plan that would work in the UA. I suggest using things like CSS 
selectors to provide a precedence order, e.g. title vs no title. This is an 
area for forward motion, but hard work. Won't fly unless we can present it 
to the author. The author has to care about it enough to articulate the 
CS and take the download hit as well.
AG perhaps 3% character hit.
CS we have hour long arguments about adding bytes of any amount.
AG interesting. we get very different reads on this issue. history with 
LINK element to provide info. but it died because authors did not use. it 
was not in the default presentation of the page. it is technically possible 
to use a DIV tree...i'm concerned that we take the theory of the structure 
and feed it to the author.
CS keyboard shortcuts and faster navigation for power users.
AG still needs to be something in the markup to drive the keyboard 
shortcuts. it takes a while to tab through stuff if you use the tab order.
JW an absolute requirement (p2 level) that appropriate markup should be 
used. we need to provide examples and make the necessity as clear as 
RS in the next release of UAAG??
JG If just techniques, put in techniques. we have other checkpoints related 
to moving to active elements.
GJR there is a responsibility of people who rely on advertising revenue to 
set up standards that the content that they pull in reformulate pages.
CS e.g. banner ads. saying it should have appropriate alt-text?
GJR yes. and the container should be marked appropriately.
AG only in the actual graph of how many hands it pushes through, it is more 
removed from chief editor of the page. my scenario: go to advertiser and ad 
agency. if they want to provide alt-text, that's cool. but to the extent 
not informative in audio, should get out of the way in audio.
GJR easy argument to make to advertiser: you're missing people. if they're 
ok with that then ok
CS there are advertisement aggregation companies. that may be where 
alt-text gets lost.
GJR suggest that to EO group. we need to raise consciousness with them.
DP that knowledge would be welcome.
@@HB suggest to the EO group that they need to raise consciousness with 
advertisement aggregation companies (that's one place where alt-text might 
be getting lost).
JW concerned that examples have all been so HTML-specific. With this next 
version we are hoping to address wider variety of technologies. What type 
of structure and semantics should be provided for different types of content.
GJR we could affect great change by using XHTML as baby steps for 
authors...it's XML with training wheels. increasing functionality of it. 
take the power of example. do a technique in HTML. then XHTML. then port 
with use of RDF to XML. we have a blueprint for people to migrate to XML.
@@GJR show what a possible blueprint migration from HTML to XHTML to XML 
might look like.
JW Yes, a great idea. One of the principal ways we can develop the 
relationship between HTML and XML to show how structure will be beneficial. 
In a generic markup language such as HTML there is a limited range of 
semantics. In this context, MAP designates an image map or a navigation 
cluster of links, generic DIV while useful to capture structure does not 
give semantics. There are ways to address it, one of which is to use an 
attribute (class) but we don't have conventions for values to use. Or use 
human-readable title. But, there are issues with capturing specific 
semantics in using general containers elements. Therefore, using XHTML 
module with XML defined in it, with a style sheet....techniques need to be 
@@WCAG WG create techniques that show how to use XML with a style sheet.
WL encouraging the use of CSS in effect causes the author to become agent 
of structure insertion.
IJ misuse of style to convey semantics. sounds like since applying style 
and adding semantics wrong way to go.
JW think the argument was that when using style sheets to convey 
presentation, a tendency to use proper markup.
IJ would like to be shown that that is the case.
AG that is a good candidate for the survey - to survey what is out there.
@@?? Is it the case that when style sheets are used to convey presentation 
that proper markup is used? Do pages that use style sheets use more 
structure markup than those that do not use style sheets?
GJR a friend filling out a tax form, did not fill out a piece of it because 
it was not in bold. people have been trained to look for style. it gives 
people scannability.
IJ your argument is that there are semantics in style and that's a good 
thing, but that you shouldn't convey semantics first through style.
AG I would not believe WL's claim based on what comes out of word. IJ - 
ideal way of doing it. making distinctions is valuable. making them in 
style sheet is more valuable.
JW there is a dependency. if proper markup is used, one can reflect the 
structure using the style sheet in a particular presentation medium. 
high-quality visual design is intended to reflect the structure and 
semantic distinctions as clearly as possible. from author perspective, 
consider that the markup captures the structure properties and style sheet 
will act upon the markup to create presentation that captures distinctions. 
then get rich rendering in the variety of platforms you provide style 
sheets for. one thing we have not addressed is to highlight the 
relationships between style and well-designed markup. there is an argument 
in using markup and adding style to make sure the distinctions are 
expressed in presentation.
GJR need to address. one of the more common uses of style sheets - to 
supplement what is there.
WL rather than beautify it.
CS because people have to support older browsers.
JW even in XML document the purpose of the SS is to provide presentation of 
semantics in the markup. one should think of dependency in that way and the 
relationship between good markup and high quality presentation.
HB more than one style sheets for clients. the SS should be generic to an 
XML application. with recursion, the style has to handle recursive objects. 
those represent significant challenges for e.g. going from braille to 
visual style sheet.
AG the SS ideally are not into the specific application but into the 
vocabulary of app widgets - paragraphs or toolbars. i thought that one of 
the things you said was the idea that, "who are designing people so that it 
shows up well is radically different media?" people are getting more 
interested with phone stuff in market.
WL that's why a "C" in CSS.
AG perhaps another one for harvey. i'm interested in WAI getting as smart 
as can about industry. where find people practicing what we preach.
JG I have some immediate concerns. Not suggesting that any guidelines be 
changed, but I am hearing about bridges to new technologies, are there 
things that we can use beside MAP to markup this info? how long to 
resolution? 3 months or so?
JW already today, the use of container elements and structural elements is 
important practice. guidelines and techniques in WCAG already suggest how 
to do that. at the block level that make up macro structure of document...a 
generic markup language does not provide semantics. need to use generic div 
elements. use title or class to provide semantics. in XML we have more 
options. I would be interested to where were people think more work needs 
to be done and what issues should be settled soon.
AG there are 2 examples that have constituencies in author population. in 
ISO, they require that headers have to be nested properly. if there is an 
HTML page that claims it is ISO-HTML you can apply their algorithm and say 
these are structural divisions. Similarly, there are few people converting 
from latex that use divs. I've been reluctant to say that UA should try to 
get UAs in large to do this. It would only be beneficial for small 
percentage of page problems.
CS but if in the UAs more authors will use it.
AG yes, definite chicken and egg philosophy. but want mock-ups to show 
authors to sell them.
@@WC find examples of ISO-HTML pages that are conformant (i.e., have 
properly nested headers).
@@WC check out ISO-HTML algorithm to determine structural divisions. take 
to ER WG.
JW that's authoring tools, eh?
IJ am I hearing Jon say, "tell me WCAG techniques to help UAs identify 
navigable structures."
JG yes.
IJ there are 3 classes of things i can think of:
known according to spec that meet our needs (headers)
twist with some reliability (map)
everything else (div and class) poor man's way to go. correct way is with 
schemas and xml. twisting html any more is not the right thing to do.
IJ it is useful to identify key components and what can use legitimately. 
against twisting. want to make sure schemas meet the needs of these groups.
JW reaction to technique that says, "enclose sections of doc in a container 
element with div and class"
IJ don't want to push to checkpoint level in UA.
JG not talking about checkpoints in UA.
IJ if in techniques, there's no conformance with that technique so what 
will be the impact on UA?
RS anything in WCAG about your suggestion?
AG it's there in generic language. it says "structure you page well."
IJ I have been against "class=navbar"
WL this is something for people making UAs.
IJ solution is not to give them specific cases to deal with but language to 
do that properly. still waiting for schemas.
GJR modularized techniques as baby-steps to HTML (not optimal)..
IJ how get any browsers to support it?
RS i don't like hacks because they break. i'd rather have a solution that 
you know will work. let's work on a second one, after we get this out that 
talks about XML solutions.
WL browser won't do this...maybe a screen reader will....regular UAs never 
CS power users.
JW if there is a requirement to traverse tree structure that will pick up 
any container elements irrespective of the semantics. reading of labels on 
them is up to the UA. the legacy DIV element would be supported 
automatically as far as the UA is concerned. then, appropriate authoring 
practices to use it. agree that generic elements such as DIV with class 
attribute that has no agreed upon usage is open-ended and no way to develop 
common usage patterns.
WL whatever happened to object?
IJ not well supported. not in IE5, NS6, Opera4.
AG report from HTML WG is that Object is viewed as a camel. no one was 
happy with it.
GJR if we want to retain OBJECT and its power, we need to strip out 
attributes that were thrown in to move forward.
AG that is continuing a thread in PF.
JW continue that in PF.
IJ seems to me that the topic that has been in UA for a while, "how should 
the UA provide useful navigation and what can we get out of content to 
identify them?" there have been many discussions about what is available in 
markup and it seems that we end up saying, "there is limit to what we can 
rely on in the spec and we need schemas." the dependency fell away and the 
issue became more in the realm of UAs. given markup what functionality can 
we provide. "allow the user to navigate easily to important components..." 
but requires nothing special from authors. AG pointed out that just 
providing nav mechanism based on spec is not sufficient. have not heard any 
more about what authors can do to help UAs pick up on things. not sure we 
need additional techniques.
AG what authors can do: go back to the amazon rewrite.
@@WCAG WG review AG's amazon rewrite.
IJ i have not heard any new information about what authors can do. we had a 
victory with map for navigation mechanisms. have people identified other 
components that we can identify with reliability with standard HTML. if 
not, should we focus on ...
AG back to headers, forms, etc.
IJ yes, but that is not beyond what HTML specifies. if we are saying to UA 
and authors "use specs" then nothing else to say.
AG even tho map was designated for image maps...
IJ let me get to it easily, let me skip over it, let me delve in.
JW we know that the authoring techniques: use markup according to spec, use 
structural markup (e.g., DIV), use more specific semantics in markup. on UA 
side: skip particular elements, read elements... if I were to suggest that 
UA were to navigate through tree would it be appropriate to achieve this? 
when there are relationships in markup (siblings but not in container 
elements) bring doc tree in synch with semantic tree, some way to determine 
structure with transformations?
AG can you focus on one question at a time?
IJ 1st question: what is the minimal requirement for UAs to provide 
structural navigation?
JW not min. req. but would the option of traversing the doc tree be 
appropriate way from user's perspective and UA developer to satisfy the 
RS use DOM as path to get element. within that path the user should have 
option to nav by major structures. they have to be able to access all 
elements in the document. as simple as that. following dom path to get to 
elements. use significant elements to skip over details.
WL not just headers - everything.
Cs add accesskey to divs, spans, etc.
GJR can add to : anchors, forms, etc.
CS on DIV, does not do anything. could solve problem.
JW a kludge.
CS why not allowed.
JG can not take focus.
CS can take tabindex. it highlights. in IE5.
IJ working draft: user interface css3. some are pushed to this document. 
with the "enabled" class you should be able to say that something belongs 
to tab order.
JG /* summarizes */ not much more we can do with HTML, schemas give us more 
@@GJR re-ping glen gordon and mark hakkinen to find out what they are doing 
to skip over repeated text. could be simple comparison. good for techniques 
Next WCAG WG meeting
/* many will be at ER meeting next week */
JW will have meeting next week, and whoever can make it will.

$Date: 2000/05/05 11:47:33 $ Wendy Chisholm
wendy a chisholm
world wide web consortium
web accessibility initiative
madison, wi usa
tel: +1 608 663 6346
Received on Friday, 5 May 2000 07:59:13 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:49:26 UTC