Archive index

A11y Slackers Gitter Channel Archive 19th of August 2015

What fresh hell is THIS now? - Patrick Lauke
  1. zakim-robot
    Aug 19 00:00
    [dsturley, a11y] I don’t think VoiceOver explicitly has that feature but there are apps out there that do it
  2. zakim-robot
    Aug 19 00:00
    [dsturley, a11y] https://github.com/keycastr/keycastr
  3. zakim-robot
    Aug 19 00:01
    [dsturley, a11y] some screencasting software has options for it too
  4. zakim-robot
    Aug 19 00:36
    [Rick Brown, a11y] i was reading the other day about ascii art in HTML, and how to make it accessible with <figure> and <figcaption>
  5. zakim-robot
    Aug 19 00:37
    [Rick Brown, a11y] then i was reading about including images in tweets and providing alt text for them in the same tweet or a follow-up tweet
  6. zakim-robot
    Aug 19 00:38
    [Rick Brown, a11y] but i'm wondering if anyone has any ideas on ascii art in tweets. i'm thinking it would be pretty annoying to a screen reader user?
  7. zakim-robot
    Aug 19 00:38
    [Rick Brown, a11y] could use the same alt text mechanism as for an image
  8. zakim-robot
    Aug 19 00:38
    [Rick Brown, a11y] but i think the ascii art would be read character by character
  9. zakim-robot
    Aug 19 00:39
    [Rick Brown, a11y] i think the answer is there is no answer, but throwing it out here
  10. zakim-robot
    Aug 19 00:45
    [Rick Brown, a11y] maybe nobody does ascii art in tweets and it's a moot point
  11. zakim-robot
    Aug 19 00:50
    [Rick Brown, a11y] some context: https://github.com/ricksbrown/cowsay
    Thinking about giving it an HTML output mode that produces accessible HTML but also maybe "tweet" mode.
  12. LjWatson
    10:29
    This rocks... bookmarklet tool for visually showing ARIA and ARIA interactions http://whatsock.com/training/matrices/visual-aria.htm
  13. StommePoes
    10:29
    I don't have any bookmarking anything showing up when I right-click, but then I've never used such a thing and have no idea how bookmarking/favourites work.
  14. StommePoes
    10:29
    Can this work in some other way? Like a plugin, maybe?
  15. LjWatson
    10:31
    You need to add the bookmarklet to your bookmarks/favourites or whatever the equivalent in your browser is. Then when you're on a page, go to your Bookmarks/Favourites/whatever and select the tool.
  16. LjWatson
    10:31
    I should think a browser specific plugin could be created to do the job though.
  17. StommePoes
    10:56
    I'll just have to look up how to do all that Bookmarks/Favourites/whatever stuff. Last time I had bookmarks, I was running Firefox 2.
  18. aardrian
    14:27
    @artisticabode @garcialo @heidi Just saw your video convo. I forked a pen on CodePen that uses <video> to do "background" video and just added the controls back in to make it more accessible: http://codepen.io/aardrian/pen/jPJKLg
  19. aardrian
    14:28
    @artisticabode @garcialo @heidi No idea if that is useful, but, well, yeah. There is also a CSS-only one on CodePen which I haven't taken the time to fork yet. Also cannot find URL right now...
  20. aardrian
    14:36
    Dear anyone who can answer a question about role="search" (which requires some set-up)...
  21. aardrian
    14:37
    HTML validator tells me not to use role="search" on a <form> (it's a warning).
  22. aardrian
    14:37
    Example 4 here shows <form role="search">. http://www.w3.org/TR/WCAG20-TECHS/ARIA11.html
  23. aardrian
    14:38
    Spec says any role allowed on <form>: http://www.w3.org/TR/html/forms.html#the-form-element
  24. aardrian
    14:38
    So, is it safe to ignore the warning, or should I only put a role="search" on a container of the <form>?
  25. LjWatson
    14:42
    @aardrian I wouldn't ignore the warning. If you put role="search" on <form> it will override the semantics of the form.
  26. yatil
    14:44
    I also think it is a bug in ARIA11 and in specs and not putting other roles on <form> seems to be the proper thing to do. I have asked my collegue Michael Cooper about the origins, this seems to be a glitch in the ARIA matrix :-)
  27. aardrian
    14:45
    @LjWatson Then I think the example is wrong or needs to include the warning.
  28. LjWatson
    14:45
    @Yatl yes, I think you and/or Michael is right about it being a glitch in the ARIA matrix :)
  29. LjWatson
    14:46
    @aardrian yes, areed. I'd also suggest filing an issue against the WCAG spec.
  30. aardrian
    14:47
    @LjWatson Sadly, I don't have the AT experience to see the impact. I had assumed that landmark roles didn't blow up other semantics.
  31. LjWatson
    14:47
    @Yatl do you know the best way to file issues against WCAG techniques?
  32. aardrian
    14:48
    @LjWatson I can file a bug. I'm happy to re-learn that.
  33. LjWatson
    14:49
    This W3C note has more info on why overriding native semantics isn't a good idea in general: http://www.w3.org/TR/aria-in-html/
  34. cptvitamin
    14:50
    good find @aardrian that example in aria11 is definitely incorrect
  35. LjWatson
    14:50
    There are times when it can (and even should) be done. I don't think putting search on <form> is one though. Better to put the landmark on a <div> containing the form.
  36. aardrian
    14:50
    @LjWatson Yeah, I proudly skip wrappers when I can, so I'll need to go rebuild that.
  37. LjWatson
    14:51
    It might not have much real-world impact (without testing I can't be sure), but telling an AT that a form is no longer a form sounds like a recipe for disaster.
  38. LjWatson
    14:51
    Yeah, I'm generally with you on the wrappers thing as a rule.
  39. aardrian
    14:51
    @cptvitamin Thanks. Now I gotta figure out how to file a bug/issue against as I cannot edit.
  40. cptvitamin
    14:51
    @LjWatson or the title of a Magritte painting
  41. yatil
    14:52
    This github repository is the best place to find bugs, it is closely monitored by WCAG WG: https://github.com/w3c/wcag/
  42. aardrian
    14:52
    @LjWatson If you ever do test, I'd love to hear results. Again, my brain separates landmark roles into their own bucket, so I always assumed other semantics were retained.
  43. cptvitamin
    14:52
    ce ne sont pas une forme
  44. LjWatson
    14:53
    Thanks @Yatl
  45. yatil
    14:55
    The thing that is weird about this is that the validator is rarely complaining about ARIA things but very vocal on not adding a role to forms while that seems not to be backed by neither the HTML5 spec nor the ARIA spec. (Although I, personally, think it is the right thing to do.)
  46. LjWatson
    14:55
    @aardrian where in the ARIA 1.1 draft do you see that example?
  47. yatil
    14:55
    @LjWatson It is ARIA11, the WCAG technique: http://www.w3.org/TR/WCAG20-TECHS/ARIA11.html#ARIA11-ex4
  48. LjWatson
    14:56
    @Yatl Aha! Thanks :)
  49. aardrian
    14:57
    @LjWatson Yeah, @Yatil is the champ who found that.
  50. LjWatson
    14:57
    Agree about the conformance checker. Suspect the hand of @sideshowbarker and/or @stevefaulkner there
  51. LjWatson
    14:57
    @aardrian you ok to file an issue? Can easily do so if you'd rather not...
  52. aardrian
    14:58
    @LjWatson I am so confused. I am wandering around a Git repository. I feel like I should just be filing a bug in the Bugzilla.
  53. yatil
    14:58
    @aardrian Click New issue here: https://github.com/w3c/wcag/issues/
  54. yatil
    14:58
    (top right)
  55. aardrian
    14:59
    @yatil Ah. I'll give that a go.
  56. aardrian
    14:59
    Nice to know this will be archived so everyone can see my cluelessness.
  57. zakim-robot
    15:01
    [Karl Groves, a11y] Don’t worry @aardrian I also took a screenshot
  58. aardrian
    15:01
    Karl, you're a wanker.
  59. zakim-robot
    15:03
    [Karl Groves, a11y] :broken_heart:
  60. aardrian
    15:03
    Ok, my attempt at a bug report: w3c/wcag#113
  61. cptvitamin
    15:07
    Not to complicate matters even more but the form role as specified in WAI-ARIA 1 says "For search facilities, authors SHOULD use the search role and not the generic form role. "
  62. cptvitamin
    15:08
    http://www.w3.org/TR/wai-aria/roles#form
  63. yatil
    15:08
    Thanks @aardrian.
  64. cptvitamin
    15:08
    But the document conformance section of ARIA in HTML says that an author may not use any other role on the form element http://www.w3.org/TR/html-aria/#document-conformance-requirements-for-use-of-aria-attributes-in-html
  65. yatil
    15:08
    And NOOOOOOO! @cptvitamin :-D That is interesting…
  66. aardrian
    15:09
    @cptvitamin Why do you hate us?
  67. jonathantneal
    15:09
    Is there a guide online to setting up AT testing environments, similar to setting up browser testing environments?
  68. cptvitamin
    15:09
    this is actually how I show love
  69. LjWatson
    15:10
    The difference is that the ARIA spec is talking about using the role on a semantically neutral element like a div, where as the HTML spec is talking about overriding the native semantics of an existing element.
  70. aardrian
    15:10
    @cptvitamin You and I might be a fit after all.
  71. LjWatson
    15:11
    So in theory <div role="search">//Form fields here</div> is correct, and so is <div role="search"><form>//Form fields here</form></div> but not <form role="search">//form fields here</form>
  72. cptvitamin
    15:11
    I was thinking that too @LjWatson but I can see this being very confusing to many
  73. LjWatson
    15:11
    @cptvitamin hellishly confusing, yes.
  74. cptvitamin
    15:11
    @LjWatson agreed
  75. aardrian
    15:12
    So...
  76. aardrian
    15:13
    Was that bug I just filed premature? Because my brain broke a moment ago.
  77. aardrian
    15:13
    Strike that.
  78. zakim-robot
    15:14
    [Karl Groves, a11y] I blame the W3C
  79. LjWatson
    15:22
    @aardrian no, it's fine. I've just added some replacement content for the technique.
  80. aardrian
    15:26
    @LjWatson Ah. I see now that I could have written that bug report much betterly.
  81. LjWatson
    15:27
    @aardrian the bug report was absolutely fine.
  82. aardrian
    15:28
    @LjWatson Thanks. That's lovely of you to say.
  83. yatil
    15:33
    Looks like we need to file that against WAI-ARIA & PFWG. I will take care of that to present the whole discussion we had. Thanks for all the input!
  84. stevefaulkner
    15:41

    A11ySlackers chat log 18th August 2015 http://w3c.github.io/a11ySlackers/archives/2015/08/2015-08-18.html

  85. aardrian
    15:50
    @stevefaulkner Your text is so large, for what are you compensating?
  86. stevefaulkner
    15:50
    @LjWatson the conformance requirements for role use on <form> have been tightened if people no likey and can provide reasoning for why it should not be so then please file a bug on http://www.w3.org/TR/html-aria/
  87. stevefaulkner
    15:52
    @aardrian for your use of 'RT'
  88. aardrian
    15:53
    @stevefaulkner Fair.
  89. yatil
    15:54
    @stevefaulkner When and where did that tightening happen, we were unable to find it in the specs…
  90. stevefaulkner
    15:56
    When I split out the conformance criteria for aria use in html into a module spec of 5.1
  91. yatil
    15:56
    So that is an editor’s draft? (I wonder as it is in the validator…)
  92. LjWatson
    15:58
    @stevefaulkner no argument from me - there is no good reason to override the form semantics in this case that I can see.
  93. stevefaulkner
    15:59
    @yatil as far as when conformance requirements are implemented in the validator that's @sideshowbarker's call
  94. LjWatson
    16:00
    @stevefaulkner is this the latest version of ARIA in HTML module? https://specs.webplatform.org/html-aria/webspecs/master/
  95. yatil @yatil agrees with @LjWatson
  96. dylanb
    16:03
    @stevefaulkner ARIA needs the ability to specify multiple roles (from different categories). Landmark and widget roles should be allowable at the same time on the same element. I know this probably totally blows the accessibility API capabilities out of the water, but it would make sense
  97. LjWatson
    16:04
    Thnks @Yatil :)
  98. aardrian
    16:04
    @dylanb "Like."
  99. yatil
    16:04
    role="search form" @dylanb?
  100. dylanb
    16:05
    @yatil yes
  101. aardrian
    16:05
    @yatil Also "like."
  102. LjWatson
    16:06
    Possibly... but role="form" is a landmark.
  103. dylanb
    16:07
    yes, you should also be able to have two landmark roles if that makes sense
  104. LjWatson
    16:07
    Which is odd in itself, since that isn't mapped through to <form>
  105. dylanb
    16:07
    another good example is role="menubar navigation"
  106. yatil
    16:07
    <button role="heading button"> mind blown ;-)
  107. LjWatson
    16:07
    No, it doesn't make sense :)
  108. dylanb
    16:08
    what does make sense is role="tab heading"
  109. LjWatson
    16:08
    At least not generally. This use case might be the only exception.
  110. LjWatson
    16:08
    I can't think of any other combinations of landmarks that would make sense...
  111. dylanb
    16:08
    there are a bunch of combinations of structure roles and widget roles that make sense
  112. dylanb
    16:09
    @LjWatson yes, that is rarer
  113. dylanb
    16:09
    but search could be fixed by superclassing it off of form
  114. LjWatson
    16:09
    @dylanb I think that combining tab and heading is because we're so used to using headings (as screen reader users) as a form of navigation.
  115. LjWatson
    16:10
    Whereas if screen readers had a similar shortcut for navigating to/between tabs, the heading part would become redundant in a tab.
  116. LjWatson
    16:10
    es, superclassing search/form would make sense.
  117. LjWatson
    16:10
    *yes
  118. aardrian
    16:11
    If landmark roles are primarily for navigation, then do they impart more semantics than necessary for that purpose?
  119. aardrian
    16:11
    Can't they pair with document structure roles pretty easily?
  120. dylanb
    16:12
    @LjWatson yes, but you could still make the argument that role="navigation" could be applied to many different widgets (tabs, menus, trees)
  121. dylanb
    16:12
    potentially even grids
  122. LjWatson
    16:20
    Yes, but the thing itself isn't the navigation. Landmarks are the regions of the page containing those htings.
  123. LjWatson
    16:21
    Landmarks represent the high level chunks of the page. You might have two types of content (tree/list of links) in a single navigation landmark region.
  124. LjWatson
    16:22
    I suppose you could be told there was a nav-tree followed by a nav-list, but it could get horribly granular I think.
  125. LjWatson
    16:23
    It's worth thinking about a lot more though. These are just first impressions off the top of my head...
  126. dylanb
    17:03
    Currently, we are forcing developers to place (potentially extra) tags around structures just to facilitate the landmarks.
  127. stevefaulkner
    17:12
    @dylanb a look at any but showsite code indicates that there are many many elements used that could be candidates for role addition
  128. stevefaulkner
    17:13
    It is a design decision of ARIA that roles always override - not saying good/bad, just is
  129. stevefaulkner
    17:14
    My reasoning for restricting form is that it has a strong native semantic
  130. powrsurg
    17:18
    What is the logic behind role="none"?
  131. dylanb
    17:18
    Now that we have some experience with ARIA, I think we should open up some of these topics for discussion
  132. dylanb
    17:18
    @powrsurg don't ask
  133. powrsurg
    17:19
    I thought it was weird that status.modern.ie had that even listed separately from ARIA roles for support
  134. dylanb
    17:20
    it is a synonym for role="presentation" and [deleted rant]
  135. StommePoes
    17:45
    The only reason I wrap stuff is often purely for roles, or role-containing elements. <nav> I only add for the roles, it's pimping my ul's
  136. dylanb
    18:01
    What he said :point_up: August 19, 2015 1:45 PM
  137. powrsurg
    18:24
    does anyone actually use the built-in Windows screen reader?
  138. zakim-robot
    18:27
    [Marcy Sutton, a11y] I think using a wrapping element for a landmark role is a fine cost to reduce the complexity required to combine roles. i.e. putting a landmark on a widget. Which role takes precedence? Do they join forces? A lot of complexity to figure out.
  139. StommePoes
    18:35
    sometimes we're scholars and poets
  140. StommePoes
    18:35
    A search widget, a navigation panel.
  141. StommePoes
    18:35
    Not that I'm arguing for combining roles on things, but it's an interesting idea.
  142. StommePoes
    18:36
    In any case, once someone decides you must schema.org-ize ALL The Things, trying to keep markup (c)lean is sorta pointless.
  143. zakim-robot
    19:03
    [Jesse Beach, a11y] @marcysutton only one role can be active on an element at a time. Although I think I misinterpreted your statement
  144. zakim-robot
    19:03
    [Jesse Beach, a11y] Can you post a gist?
  145. zakim-robot
    19:09
    [Jesse Beach, a11y] I long ago gave up the battle to produce clean HTML. 3 years in Drupal beat the need out of me. Ultimately, it's just painting the inside of the bike shed.
  146. StommePoes
    19:10
    ug, drupal. I keep hearing drupal 8 won't have the 500 nesting divs thing
  147. zakim-robot
    19:10
    [Jesse Beach, a11y] As long as it's not needlessly bloated (empty divs wrapping empty divs down to abysmal depths), then meh.
  148. StommePoes
    19:10
    Jesse, marcy was replying to earlier discussion about what-if multiple roles, rather than her thinking we are currently allowed multiple roles
  149. zakim-robot
    19:11
    [Jesse Beach, a11y] @StommePoes My opinion is it produces as much divery as it needs, but not much more
  150. StommePoes
    19:11
    block block-block block-leaf leaf-block block-leaf-block block-block-leaf duck duck-duck duck-duck-goose
  151. zakim-robot
    19:11
    [Jesse Beach, a11y] @StommePoes, gotcha, I thought I must be misinterpreting the comment :simple_smile: thanks for clarifying
  152. zakim-robot
    19:11
    [Jesse Beach, a11y] ahaha
  153. StommePoes
    19:11
    it makes way more than necessary for simple things. I've crawled through megamenus built in drupal and it was eye-bleaching
  154. zakim-robot
    19:12
    [Jesse Beach, a11y] ya, well, that's the tradeoff for a page building framework :simple_smile:
  155. StommePoes
    19:12
    I could have removed a good 10 layers without changing anything that was actually necessary
  156. zakim-robot
    19:12
    [Jesse Beach, a11y] I make no claims for contrib!
  157. StommePoes
    19:12
    yeah, understood
  158. StommePoes
    19:12
    But again, I keep hearing 8 won't do this, or anyway way way less
  159. zakim-robot
    19:13
    [Jesse Beach, a11y] But I worked on the drupal core render method, cacheing, whatnot. That's really a commendable bit of code
  160. StommePoes
    19:13
    um, heard from a danish guy, trying to think of his name... wore flat hats and had glasses and some kind of little facial hair in the front... would recognise his name I think
  161. zakim-robot
    19:13
    [Jesse Beach, a11y] MortenDK?
  162. StommePoes
    19:13
    aha, yeah. He gave a quick lightning talk at Fronteers like 2 years ago
  163. StommePoes
    19:14
    and he showed the Perilous Tower Of Divs
  164. zakim-robot
    19:14
    [Jesse Beach, a11y] Yup, that'd be him
  165. StommePoes
    19:14
    which we all groaned at
  166. zakim-robot
    19:14
    [Jesse Beach, a11y] hahah, ya, he's a div ascetic
  167. StommePoes
    19:14
    and he promised us Good Things, and we rejoiced.
  168. StommePoes
    19:14
    ascetic? Ascetic in the land of coca-cola santa claus with extra diabetes!
  169. zakim-robot
    19:14
    [Jesse Beach, a11y] :tada:
  170. StommePoes @StommePoes claps
  171. StommePoes
    19:15
    I had bootstrap poured down my throat recently and I can't even swallow that. Ug.
  172. zakim-robot
    19:16
    [Jesse Beach, a11y] Me, I'm a bit of a framework apologist :simple_smile: Having worked a few at the core level, it's sobering to realize how hard it it to build them in a sane way
  173. StommePoes
    19:19
    there is no sane way, I don;t think
  174. StommePoes
    19:19
    that's why, like CMSes, everyone says they all suck and builds their own
  175. StommePoes
    19:19
    which then also sucks
  176. StommePoes
    19:19
    so the best ones are either very very domain specific
  177. StommePoes
    19:19
    or are extemely structured in what's allowed in core vs plugins
  178. StommePoes
    19:20
    and core tries to only address 80-20
  179. StommePoes
    19:20
    solve 80% of needs with 20% of rules/code. And yeah, that's hard
  180. StommePoes
    19:21
    It's probably easier to do the domain-specific ones
  181. StommePoes
    19:22
    so stuff like WP, Drupal, maybe even Perch, those are pretty hard to do right.
  182. StommePoes
    19:23
    It's nice that both WP and Drupal include a11y as a necessary part, makes things that much easier for developers using them
  183. zakim-robot
    20:56
    [Marcy Sutton, a11y] @jessebeach: I was responding to the “what if we could use multiple roles” argument :smile:
  184. zakim-robot
    20:57
    [Jesse Beach, a11y] @marcysutton: gotcha. As I was writing, I was thinking "Marcy certainly knows this; I'm just misunderstanding the conversation!"
  185. zakim-robot
    20:59
    [Marcy Sutton, a11y] I do like “what if” questions though, they create opportunities to potentially improve things if we really do find a need
  186. zakim-robot
    22:03
    [dylanb, a11y] @marcysutton: there is some figuring out to do for sure, but that is not an argument against it. In a new application, being developed from scratch with accessibility in mind, an extra div or two won't be a big burden. In a legacy project, it may break all sorts of code/CSS and introduce tons of "extra" work. The end result is that developers will likely leave out the landmarks (because they are not really a requirement to be accessible, they just make things nicer). Allowing multiple roles would allow them to get the added advantage of the landmark navigation with less effort.
  187. zakim-robot
    22:03
    [dylanb, a11y] </soapbox>
  188. zakim-robot
    22:04
    [dylanb, a11y] :smile:
  189. zakim-robot
    22:04
    [Marcy Sutton, a11y] we can agree to disagree on that one :smile: