User Agent Accessibility Guidelines Working Group Teleconference

09 Jan 2014

See also: IRC log


Greg_Lowney, Jim_Allan, Kim_Patch, jeanne, +1.425.381.aaaa, KFord
JimAllan, KellyFord


<trackbot> Date: 09 January 2014

<scribe> scribe: allanj

<jeanne> [discussion on report on browser penetration statistics]




W3 Resource Priorities paper review http://lists.w3.org/Archives/Public/w3c-wai-ua/2014JanMar/0000.html

gl: for @alt it is in the html already, so is downloaded before the resource.
... for other items @longdesc (separate document) would not be load until user request
... sample code does not include @alt, which is bad practice.
... should write them

ja: for all the html5 image elements, etc. they are in html and not affected.
... does this apply to child elements of the large buckets (embed, object, etc). can lazyload be applied to <param>

gl: concerned about @postpone

<Greg> The POSTPONE attribute is entirely based on when things are in visible viewport. It says "If the postpone boolean attribute has been specified on an element, the User Agent MUST not start downloading the resource associated with the element until either the bounding box of the element is inside the User Agent's interpretation of the Document's viewport or the element has been styled such...

<Greg> ...that its display property is no longer set to none."

<Greg> If you're using AT that reads from the DOM, and aren't visual and so don't need to scroll things into view, this will prevent you from reading them.

gl: screen readers don't scroll the viewport, so DOM not updated, and resource not downloaded

<Greg> How does "postpone" work with searching the page for a string, etc.? If content you haven't scrolled to isn't loaded into the DOM, search returns "no hits"?

ja: perhaps screen reader should scroll the screen as they move, and provide a focus ring

<jeanne> ja: does search trickle through an iframe?

<Greg> If search includes content of iFrame, or includes alternative text which (for things like longdesc) aren't loaded automatically with the page, then presumably they would not be searched.

<Greg> audio and video would also have their text alternatives (captions, transcripts) be external to the page, so postpone could prevent them from being read or searched until scrolled to.

gl: UAAG has SC for the ability to search alternative content. very important for audio and video content
... should UAAG have an SC to allow user to override the 'resources' attributes, to make the document download in source code order

<jeanne> JS: I think we should comment to Web Perf group that the document needs to specify that alternative text [alt, longdesc and ARIA describedby and labeledby attributes] needs to be loaded with the text and not be lazyloaded or postponed.

<jeanne> Maybe we could say that all the text attributes associated with an image, audio or video should not be lazyloaded.

need clarification that html is downloaded with all attributes, and the dom built. then the resources (images, audio, video, etc) are downloaded.

js: text elements are necessary for a11y

<jeanne> JS: Postpone elements may not be available to the accessibility APIs.

js: concern about @postpone

ja: the @postpone uses "User Agent's interpretation of the Document's viewport",
... if user loads a page and hits PRINT, then the viewport changes to the entire document, and all resources need to be downloaded, for the printing viewport

<jeanne> JS: I am concerned that an @postpone element may never meet the criteria that would allow it to be displayed to a screenreader.

ja: does a browser know where the screen reader point of regard is located, so the resource is loaded.
... we are talking about milliseconds, will the user notice

js: need another option so information is available to AT

<jeanne> JS: We certainly need another option for postpone that will make it available to a screenreader.

<jeanne> ... what would be the screenreader equivalent of scrolling into view.

<jeanne> JA: That would be a screenreader problem, because the screenreader doesn't communicate its shadow DOM to the browser.

<jeanne> KF: JAWS has that now, but it doesn't work consistently

<jeanne> JA: We still need screenreader Focus Ring\



ja: see http://www.w3.org/TR/resource-priorities/#resource-priorities-property

user could override all of this through the use of user CSS and !important

<jeanne> http://www.w3.org/TR/resource-priorities/#the-postpone-attribute

js: overriding by user style sheet needs to be in comment

<jeanne> JS: It would be important to state clearly in this spec that user stylesheets can override this spec.

gl: why is there 2 ways (html and css) to turn on these resources priorities

<jeanne> GL: They don't say which takes precedence -- CSS solution or the HTML solution

js: mobile - which doesn't support css well

<jeanne> GL: they are designed to act the same, but behave differently.

<jeanne> ... they don't even spell lazyload / lazy-load the same.

gl: lazyload (html) vs lazy-load (css)

<jeanne> ... turning off one, would not override the other.

<jeanne> ... the HTML attibutes are not part of the CSS and would not be overridden

gl: which has precedent (the css or the html attribute)

<jeanne> ... by a user stylesheet

ja: and would !important override the html.

js: what about mobile and no user CSS

gl: inpage search, the UA does not give user any feed back that the page has not finished downloading.
... elements not yet downloaded can't be searched.

ja: see Facebook and other infinite scroll implementations
... can't search, bookmark, etc.

gl: if UA is aware that the entire page has not yet been downloaded, it could/should tell the user of this fact.

<jeanne> JA: Screen magnifiers are using the browser proper.

<jeanne> JA: for speech users, the mouseless browsing links are part of the HTML and will load.

<jeanne> JA: I think the lazyload is a matter of milliseconds and we don't have to worry about it, but the @postpone could be a problem

<jeanne> GL: @postpone has no effect unless something is something is being toggled on and off with display properties.

gl: @postpone does not make anything worse for AT for elements set to display: none

If the postpone boolean attribute has been specified on an element, the User Agent MUST not start downloading the resource associated with the element until either the bounding box of the element is inside the User Agent's interpretation of the Document's viewport or the element has been styled such that its display property is no longer set to none.

<jeanne> KF: There were more problems caused when the AT didn't honor display none, then the @postpone won't cause a problem blocking display of items set to display none.

<jeanne> JA: How do we want to handle it?

js: set another conference call to finalize this.

<jeanne> ... schedule a call and turn this into helpful comments.

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.138 (CVS log)
$Date: 2014-01-09 19:38:34 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.138  of Date: 2013-04-25 13:59:11  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Found Scribe: allanj
Inferring ScribeNick: allanj
Default Present: Greg_Lowney, Jim_Allan, Kim_Patch, jeanne, +1.425.381.aaaa, KFord
Present: Greg_Lowney Jim_Allan Kim_Patch jeanne +1.425.381.aaaa KFord
Regrets: Jan
Found Date: 09 Jan 2014
Guessing minutes URL: http://www.w3.org/2014/01/09-ua-minutes.html
People with action items: 

[End of scribe.perl diagnostic output]