25 Sept 2000 minutes

available at: http://www.w3.org/WAI/ER/2000/09/25-minutes.html

25 Sept 2000 ERT WG telecon

Summary of action items and resolutions
·       Action WC and WL to review UAAG working draft last call in the 19 
October - 9 November timeframe.
·       Action LK Review 1 September 2000 UAAG version. Does it refer to 
tooltip? If so, is there a way to configure how long it is displayed? If 
not, send the group a note.
·       Action WC: Is CVS smart enough to pick up that two lines have been 
switched?
·       Action LK: post message to list to ask Kynn to describe the 
philosophy.
·       Action WL and GR: write the definitive paper on the cyber-ghetto. 
Why it matters and how it matters.
·       Resolved: Checking for equivalent alternative pages is still an 
open issue.
·       Resolved: We don't need to check that there is only one H1 per 
page. This issue is closed. The technique has already been removed from the 
AERT.
·       Action LK: Continue discussion of javascript and 
onfocus/onmouseover on the list.
·       Resolved: The issue with scripts is still open.
·       Action LK and WC: coordinate with Jutta and CMN re: joint AU/ER 
meeting next week. Postpone?

Participants
·       William Loughborough
·       Len Kasday
·       Chris Ridpath
·       Dick Brown
·       Wendy Chisholm
·       Brian Matheny
Regrets
·       Gregory Rosmaita

Need for tools
LK There may be some things that you don't need a tool for. e.g., bring up 
a page have an image with red letters on black background and unable to read.
BM Light grey on grey backgrounds, GIF that circulates betwen colors.
WL The tool can be a human.
LK Any browser and a human will do it. Previously, Chuck Opperman described 
which order alt and title were displayed (on mouseover).
WL What is title used for? Does the spec say how to render?
WC /* reads from HTML spec. */
LK Perhaps UAAG ought to specify how long to display the tooltip/title? 
Allow configuration of length of time.
Action WC and WL to review UAAG working draft last call in the 19 October - 
9 November timeframe.
Action LK Review 1 September 2000 UAAG version. Does it refer to tooltip? 
If so, is there a way to configure how long it is displayed? If not, send 
the group a note.

Testing for alternative pages
LK Keeping track of status of various elements. CMN came up with an 
interesting technique. Instead of assigning an ID to the element in the 
HTML, you could assign a checksum to the element. Even if people rearrange 
the page, in principle you can find out whether the element has been 
checked previously.
WL Is there a conceivable non-human intervention way to determine if an 
alternative page is equivalent?
LK Let's brainstorm.
WL Playing devil's advocate, say there is not and never will be.
LK But, how far can we get? Are there some tests?
WL Find those things that people have discovered where this messes up. Did 
they change the main page and not change alternative page.
LK That's another one of those that is not HTML but is something that needs 
to be done.
DB Some alternative pages are merely a different style sheet. The content 
is the same.
WL That's not what is meant by this page.
DB It has a different look and we do call it an alternative.
WL If all you change is the presentation, then the situation is a perfect 
implementation.
DB It's more or less coming out of a database, takes a couple paths. The 
table is unwrapped. Two different pages and two different sets of HTML are 
produced.
LK If you are on the server you can see the date of the file. That's a 
whole new set of techniques for the document.
DB You could be looking at the data on the server. The data is identical 
between them. But are people getting the same meaning out of these two pages?
WL That's b/c we've been looking at "lookers" for so long.
LK If I took 2 paragraphs and swapped them, the UNIX algorithm is not smart 
enough to figure that out.
DB Right.
WC CVS is a bit smarter about this.
LK Smart enough to pick up if two lines are swapped? e.g., two lines of C++ 
code.
Action WC: Is CVS smart enough to pick up that two lines have been switched?
WL We are concerned about alt pages being up to date.
WC And making sure the data is there.
LK Edapta's specialty is to make different versions of pages.
WL Because of abuse, we have discouraged it.
LK In addition to accomodating people with old screen readers, are there 
other variations of the MSN home page? e.g. for palm size?
DB You can sign up for the mobile web, that's different. They also figured 
that page would be better for people with cognitive disabilities since the 
main page is so overwhelming.
WL They are moving in the direction of a place where you can get at the 
semantics without depending on a particular browser.
LK I'm looking at the home page. You won't find a link to it yet. You can 
get to it from the accessibility site. Cynthia was still working on the 
personalization page when she left. Comparing the text would do it.
WL The way to make sure that the alt page is equivalent is to test the 
semantics.
WC e.g., as long as it's captured in RDF.
WL right.
WC What about now? How determine that have the same info on both pages? On 
one may have IMG w/alt, on other have text. Looking for the text of the alt 
may not help if they have expanded on the alt page.
LK If have a user profile, then the person gets a page adapted to them 
(this is the Edapta philosophy).
WL Not even just for disability, but like CC/PP, what I want to be bothered 
with not just what I want to reveal about myself. Then the user agent can 
be more active.
Action LK: post message to list to ask Kynn to describe the philosophy.
LK It's more than the browser, it's the server providing in various formats.
WL For our purposes, we need to find out what it is about alt. pages that 
makes them seem like a ghetto to the person for whom it feels like a ghetto.
WC Don't we know that's it date and lack of content?
WL I don't know. I don't ever use them.
LK When I asked about, I got a few responses. E.g., a parent is looking at 
a page with their child or with her friends and talking about a page. It's 
not just the facts, it's the layout and other features.
WC What info do we need to do to move forward?
WL When someone asks, "what's wrong with an alt-page" we can say, here's 
what's wrong with.
Action WL and GR: write the definitive paper on the cyber-ghetto. Why it 
matters and how it matters.
WC Then we can figure out what we're testing for?
WL Right.
LK On the algorithm side, any more ideas?
/* none */
Resolved: Checking for equivalent alternative pages is still an open issue.

Proper use of headings
LK How many H1s can there be? Can there be only 1?
WL It is suspicious.
CR Depends on the text. If > 100 words, not a heading. If 5, most likely a 
heading.
WL Even if only 5, could be heading.
CR Can have multiple h2 or h3 why not have more than one h1?
WL Not that you can't, but it is suspect.
LK The browsers relegate titles to a strip. It does not give you that 
visual punch. If you take that philosophy, the real title is the real title 
of the page it should not be h1, but regular text with css.
WL You won't have success with. The open issue is, should we delete the 
check "only one h1." Doesn't matter to accessibility. It mattters that we 
structure things.
WC Where does this come from? Not in HTML spec. Any user agent or AT use H1 
in a special way?
WL It's because of authors abusing.
WC But can't you check if it is nested correctly? H1 H2 H3 H2 H3 H1 H2...
WL I am in favor of deleting.
LK But, you say it is suspicious.
WL Strike it.
CR I agree.
LK What does WCAG say?
WC /* reads from WCAG: 3.5 Use header elements to convey document structure 
and use them according to specification. [Priority 2] For example, in HTML, 
use H2 to indicate a subsection of H1. Do not use headers for font effects. */
WL That's good enough.
WC Favors deleting as well.
DB also favors
CR Already removed.
Resolved: We don't need to check that there is only one H1 per page. This 
issue is closed. The technique has already been removed from the AERT.

Scripts
WC Talked with Raman last week about a client-side proxy to evaluate 
JavaScript. Primarily for browsers that don't support javascript.
WL Reading about XForms. Allows you to get forms done without scripting.
LK Talking about validation?
WL Yes, which you allow and etc.
LK How about when have choice ABCD and other. Other is a text field, only 
valid if you check other. Until then it is greyed out (like GUI interface). 
Will they be looking at this?
WL Not sure. Look at "Requirements for XForms."
LK Back to scripting: Mouseovers. When are they an accessibility problem 
and when are they not a problem. Just decoration, no problem, but popping 
up text is a problem.
CR For every single mouseover, just put in an onfocus. This should be required.
WL Yes, that's what matters.
CR Not a big deal to put it in.
WC There was an issue with onfocus in that the result was active forever 
after the focus. Had to pair that with onlosefocus.
LK issue also with physical disabilities? increase access time?
CR no, same as onmouseover.
LK There are cases where can't simply have onfocus which is same as 
mouseover. e.g., if you have a page with text fields, one has focus, 
someplace else has a menu. when you click that if another place where click 
w/mouse that populates field.
WC Not accessible to have focus in more than one spot.
LK There are 4 cursors in jaws.
WC Don't understand what you're saying. Does not sound accessible or user.
LK Sure, in Eudora you can start an e-mail message, then go to address book 
and click on an address that you want inserted into the new message. That 
argues for separate control. We need to think in terms of having an 
interface where you can do things as equivalent of mouse without changing 
text focus.
Action LK: Continue discussion of javascript and onfocus/onmouseover on the 
list.
Resolved: The issue with scripts is still open.

Next meeting
LK Joint meeting with AU next week.
WL A lot of us will be in Bristol.
Action LK and WC: coordinate with Jutta and CMN re: joint AU/ER meeting 
next week. Postpone?

$Date: 2000/09/25 15:41:00 $

--
wendy a chisholm
world wide web consortium
web accessibility initiative
madison, wi usa
tel: +1 608 663 6346
/--