Archive index

A11y Slackers Gitter Channel Archive 27th of January 2016

What fresh hell is THIS now? - Patrick Lauke
  1. zakim-robot
    11:48
    [joly] I am running a livestream relay of the Geneva Engage conference. I've been able to overlay CART. https://livestream.com/internetsociety2/genevaengage/
  2. zakim-robot
    11:48
    [joly] To actually participate https://diplo.adobeconnect.com/genevaengage/
  3. StommePoes
    12:03
    none of those play for me
  4. StommePoes
    12:03
    but the picture/screenshot they show before you try to mess with controls look cool
  5. jkva
    12:04
    Worked on my machine (Chrome, OSX Yosemite)
  6. StommePoes
    12:04
    Chromium Linux it just shows the controls and a grey stripey background.
  7. zakim-robot
    13:19
    [fstorr] I have a question (actually many questions) about those web-based virtual chat assistants (the kind of thing that’s usually embedded on a page and you can type a question into and it’ll either say “I don’t know what you mean” or bring back a a useful answer). There doesn’t appear to be anything immediate in WCAG that defines those, but would they be classed as time-based media? It’s not audio or video though, so probably not. The nearest thing I can find is G193 “providing help by an assistant in the web page” (https://www.w3.org/TR/WCAG20-TECHS/G193.html) but I’m not sure that really covers it as it’s specifically talking about providing information on the content of a page rather than, say, a general question the site might have an answer for.
  8. zakim-robot
    13:27
    [karlgroves] Chat itself is def. not time-based media. In many cases, it would also be considered exempt from “Timing Adjustable” (depending on use case). IMO you can think of it was any other dynamic content. It it auto-updates outside of the user’s current focus, some mechanism (a live region) should be provided to notify the user of the new content.
  9. zakim-robot
    13:34
    [fstorr] I was almost certain it wasn’t time-based, so thanks for confirming that. Looking at this http://www.uscis.gov/ as an example, there’s a live region that updates the content after a question is submitted. There’s some odd behaviour: search for “green card” and the widget updates and also navigates the main page to a new one on green cards.
  10. zakim-robot
    13:35
    [fstorr] WRT to ARIA: is this a rare instance of role=“application”? And should focus be managed so a user can’t accidentally tab out of it and back onto the page?
  11. StommePoes
    13:37
    Is there also a requirement for a method for users to silence things that auto-update on pages, or do users already have that control? (I've successfully avoided web pages with chats so far, but don't know of any way to say "turn that crap off")
  12. jkva
    13:37
    Wasn't that part of the WCAG?
  13. zakim-robot
    13:37
    [callumacrae] With VO you can press ctrl to make it stop talking, but not sure of any way to indefinitely stop it short or turning it off
  14. jkva
    13:37
    I remember seeing something about that recently
  15. StommePoes
    13:39
    With VO you can silence live regions?
  16. StommePoes
    13:42
    fstorr I managed to fall out of the chat thing because I didn't immediately see that i was focussed on the textfield
  17. StommePoes
    13:42
    An SR user wouldn't necessarily have that problem but just sighted keyboard, the focus is correctly being put in the right place but maybe it needs a more obvious than normal visual focus ring
  18. StommePoes
    13:43
    because I can see wanting to go back to the page but keep the chat open... suppose it gives me instructions to do stuff on the page?
  19. fstorr
    13:48
    @StommePoes the visual highlighting does need more work, including tabbing over the controls at the top of the widget: there's a very subtle change of shade but that's it.
  20. MichielBijl
    13:50
    @StommePoes: the ctrl key silences the current spoken thing in VO.
  21. MichielBijl
    13:50
    Works with eveyrthing.
  22. MichielBijl
    13:51
    So if something is announced double or something, or you're simply fed up with Daniel talking through the audio, you can shut him up.
  23. fstorr
    13:51
    @StommePoes @MichielBijl the problem is that it's only temporary: if you go to another page, VO starts back up again. There doesn't seem to be a "I want VO on but not talking at all" toggle; the only way I've found to do that is to quit VO.
  24. MichielBijl
    13:52
    Correct, just wanted to clarify what ctrl does :)
  25. StommePoes
    13:54
    Thanks. There are sometimes pages with some live region constantly spouting stuff and it would just be nice if you could continue going through the actual page in peace, after you've determined you really aren't interested in the whatever that's running along.
  26. StommePoes
    13:54
    Like pages where people show their twitter
  27. StommePoes
    13:55
    @fstorr yeah or even if you could finagle some kind of focusable after the text area, cause the first thing I did when I saw my focus was not on the button I clicked to open chat, was tab. And that tab brought me out of the chat, so I had to shift-tab back into it.
  28. StommePoes
    13:55
    For repeat-offenders, once they figure out how it works it's not bad.
  29. StommePoes
    13:55
    It's first-timers who could miss it.
  30. StommePoes
    13:56
    I had to bring my laptop into the office to have the IT guys look at why Windows Updates were being blocked.
  31. StommePoes
    13:56
    So every login, Narrator starts up because I guess once Windows sees you're using a screen reader it just sets that to ON
  32. StommePoes
    13:57
    the IT guy was like "wow this is annoying" so I said "well just turn off the sound then"
  33. StommePoes
    13:57
    Was good enough for him :P
  34. StommePoes
    13:58
    ...and I don't think role=application would work great, wouldn't that get in the way of reading what "Emma" says? or if you had to check her spelling of a word (funky gov't acronyms FTW)?
  35. zakim-robot
    14:09
    [jdan] Any tips for providing alt-text to incredibly data-dense images? my coworker is placing a "Codon table" in an article and wants to know what alt text to provide. The idea case may be an actual <table> but I believe we're limited to an image at the moment
  36. zakim-robot
    14:09
    [jdan] A "Codon table" with 4 rows, 4 columns, where each cell contains a significant amount of data
  37. WilcoFiers
    14:10
    That's quite a beast
  38. jkva
    14:10
    aria-describedby, with a hidden table?
  39. fstorr
    14:10
    @StommePoes I haven't played with role=application to know more than what I've read about it, which is scant. Re-reading https://www.marcozehe.de/2012/02/06/if-you-use-the-wai-aria-role-application-please-do-so-wisely/ it seems that it's not suitable
  40. zakim-robot
    14:10
    [jdan] @jkva I don't believe it will be possible to put a table on this page, given our limited CMS
  41. WilcoFiers
    14:11
    jdan, what does the red text in this table do?
  42. zakim-robot
    14:13
    [jdan] @WilcoFiers I am not sure!
  43. zakim-robot
    14:15
    [jdan] I usually look to the College Board for alt text inspiration :simple_smile: this one is my personal favorite https://collegereadiness.collegeboard.org/sample-questions/math/calculator-permitted/7
  44. zakim-robot
    14:15
    [jdan] <img alt="The figure presents the graph of a circle, a parabola, and a line in the xy plane. The horizontal axis is labeled x, the vertical axis is labeled y, and the origin is labeled O. The integers negative 3 through 3 appear on both axes. The circle has its center at the origin and radius of about two point two. The parabola has its vertex on the y-axis at negative 3 and opens up. The circle and parabola intersect at four points, of which two are below the x-axis and two are above the x-axis. Of the two points of intersection below the x-axis, one is to the left of the y-axis and one is to the right of the y-axis. Of the two points of intersection above the x-axis, one is to the left of the y-axis and one is to the right of the y-axis. The line slants upward and to the right, and passes through two of the four points of intersection where the circle and parabola meet, one below the x-axis and to the left of the y-axis, and one above the x-axis to the right of the y-axis. In other words, the three graphs intersect at two points." src="/sites/default/files/m-c-mc-7-3equationgraph.png_resize.png">
  45. jkva
    14:17
    I wonder if any browsers bork on an alt that long
  46. iandevlin
    14:18
    An older IE probably.
  47. StommePoes
    14:27
    someone had a 100-char limit, I thought
  48. StommePoes
    14:27
    or 130...
  49. StommePoes
    14:29
    @fstorr an example of where I did have to use role-application was on a drag and drop... according to specs, using lists with list-items ought to let the SR know to go ahead and pass on keys to Javascript, but IE wasn't doing that, so JAWS users couldn't interact. So added role-application to each list and then the list with actual text in it, those each needed role-documents around them. Messy and gross and hopefully can be removed in the future with either JAWS or IE fixes.
  50. garcialo
    14:33
    Don't think the browsers have problems with long alts.
  51. garcialo
    14:34
    @StommePoes I've only ever heard of best practice is to keep alt shorter than one hundred and some characters because of how screen readers (don't know about brailled displays) treat them
  52. MichielBijl
    14:37
    if an alt text needs te be very long it ought to be it's own text (paragraphs and what not). Or link it with longdesc.
  53. MichielBijl
    14:37
    Lot of people hate longdesc thought…
  54. StommePoes
    15:11
    longdesc has issues, but the idea of maybe linking to an alternative presenation of data-dense visual things isn't bad. it just has some pitfalls.
  55. StommePoes
    15:11
    @garcialo maybe it was an SR rather than a browser who had an internal character limit.
  56. powrsurg
    15:20
    hey, they say a picture is worth a thousand words. That alt is only 203
  57. powrsurg
    15:23
    I've been doing web dev for 14 years, I cared about accessibility that whole time, but only really pushed it hardcore in the last year or so. That said, in all that time I have never understood longdesc -- or how to test it. Going to another page to receive the description seems weird because the page itself could have its own issues and need a description there ...
  58. powrsurg
    15:23
    maybe I'm just clueless in that regard
  59. powrsurg
    15:23
    This message was deleted
  60. StommePoes
    15:48
    The idea behind longdesc is, sometimes you do have a picture worth a thousand words, and yet is generally considered visually-conveying all that (think complex charts and graphs). With Longdesc, you supply a URL to someplace (could be an in-page link I guess) which has all that description, in text, because it's too long to be appropriate for alt.
  61. StommePoes
    15:49
    Since support has sucked in a lot of places, and because people kept misunderstanding how to use it, debate raged over whether it should die or not. A lot of us just decided that, if we can't have the descriptive text visible on the page with the other text, we'd just follow Longdesc's example and have a link, often in the image caption, pointing to the long-text-description.
  62. StommePoes
    15:49
    Ideally this would then be an in-page link, rather than a separate page, to help keep that info updated.
  63. StommePoes
    15:50
    Sometimes you could have that link point to a hidden element who appears onscreen with :target for example... all sorts of little tricks can do nice things.
  64. deborahgu
    15:54
    and cultural tropes have popped up in social media networks, where longdesc is often unavailable (and sometimes alt), to just do that image description under the text in brackets. Which is odd but bettr than nothing. Longdesc is best where it's the only alternative, where you can't have a link or the like.
  65. StommePoes
    15:56
    yeah, I find links in general more accessible but sometimes the graphics guy is like, no.
  66. jnurthen
    15:56
    We use longdesc along with a textual link in our documentation. See https://docs.oracle.com/cd/B10501_01/server.920/a96521/ds_concepts.htm for an example (although I don't like the fact they used the image filename in this example - that isn't normal)
  67. StommePoes
    15:58
    you mean the first image?
  68. StommePoes
    15:58
    Honestly, that could fit into an alt and be a figure caption.
  69. jnurthen
    15:58
    It could - but others are longer
  70. StommePoes
    15:58
    Maybe it's like the sad donut, I expected m0ar
  71. StommePoes
    15:58
    ah ok
  72. StommePoes
    15:58
    It's at least clear: here's an image, here's a link to a description of that image.
  73. StommePoes
    15:59
    Someone on my work hipchat room was gushing over a particular feature of OracleDB
  74. jnurthen
    15:59
    all diagrams must have a long description in our doc. Sometimes all it is is "this image is described in surrounding text" but they all have something
  75. jnurthen
    15:59
    that is always good to hear ;)
  76. jnurthen
    16:00
    anyway - I have to drive to the evil empire now
  77. StommePoes
    16:00
    haha
  78. StommePoes
    16:00
    me as well
  79. deborahgu
    16:07
    StomePoes, yeah, when you're dealing with ebooks, adding an extra link is likely to be a big no no
  80. stevefaulkner
    16:12
    2.4 How long should a text alternative be? https://www.w3.org/TR/2014/WD-html-alt-techniques-20141023/#m5
  81. stevefaulkner
    16:15
    note: i did write that so all my biases are within
  82. jnurthen
    17:23
    @deborahgu depends on the ebook. For technical ones such as product manuals I'm not sure anyone is going to object to a visible link. Product documentation is IMO an ebook and we don't get any pushback - it is just a requirement. Another (better) example here http://docs.oracle.com/database/121/TDPPT/tdppt_realtime.htm#CBBEFIFH
  83. deborahgu
    17:24
    yes, but it's certainly a common use case in publishing that the book designer is adamant about visible link.
  84. jnurthen
    17:26
    Maybe we need to seek to educate them that the information is not just for blind users but can have other use cases too. If they are really adamant then you could have a link that only appears on keyboard focus much like skip nav links often only appear when tabbed to.
  85. jnurthen
    17:27
    personally I don't like the aria-describedby approach with hidden text in the page because that causes a lot of extra text for a screen reader user to skip past when they are NOT interested in the content of the image
  86. fstorr
    17:47
    @StommePoes thanks for that example :)
  87. deborahgu
    18:08
    low vision users and people with visual processing disorders and some cognitively impaired users, for a very short list of some of hte non-screenreader users who like long descriptions
  88. deborahgu
    18:10
    this could all be fixed by user agents having settings which expose longdesc (or ARIA) on demand, instead of assuming only people with assistive tech could ever want it.
  89. zakim-robot
    18:16
    [jessebeach] @fstorr: I'd probably make the chat window a role dialog so that it has it's own top and bottom contexts.
  90. jnurthen
    18:20
    @jessebeach the problem with role dialog is that some AT will trap you in the chat window
  91. jnurthen
    18:20
    and sometimes you want to be able to interact with the page and hop back to the chat
  92. zakim-robot
    18:22
    [jessebeach] hey @jnurthen! I wasn't aware of that. Which ATs are doing this? We have a bunch of dialogs in our UI that act as complex flyouts and we've not experienced this with e.g. JAWS, NVDA, VO, ChromeVox. I want to make sure we're not trapping folks on other ATs!
  93. jnurthen
    18:33
    yours are probably ok though right.... a user can "close" the dialog and then reopen it and go back to it later
  94. zakim-robot
    18:33
    [thecristen] Thanks to everyone who replied to me yesterday, that’s really helpful. BTW I’m more looking to set some ground rules/guidelines rather than just informing about accessibility.
  95. jnurthen
    18:33
    and it is in the same behaviour... most chat apps aren't like that
  96. jnurthen
    18:34
    what we really need is a way of differentiating modal and non-modal dialogs... really frustrating that that still hasn't happened
  97. jnurthen
    18:34
    @thecristen for ground rules I recommend getting a statement from someone as high up in the company as you can that people have to do accessibility
  98. zakim-robot
    18:35
    [cameron] targeting ARIA selectors for CSS to visually enforce correct widget markup :heart: http://whatsock.com/training/#hd11
  99. zakim-robot
    18:35
    [cameron] I love this idea
  100. jnurthen
    18:35
    we have a statement from one of our CEOs which can help
  101. zakim-robot
    18:35
    [cameron] anyone know of CSS frameworks that target ARIA selectors to prevent regressions?
  102. zakim-robot
    18:36
    [cameron] or that use ARIA selectors at all?
  103. jnurthen
    18:37
    @jessebeach - maybe i'm mistaken. If you have tested with that many ATs i'm sure it is ok. Maybe I am thinking historically.
  104. zakim-robot
    18:49
    [cameron] there's also http://www.ebaytechblog.com/2015/11/04/how-our-css-framework-helps-enforce-accessibility/
  105. zakim-robot
    18:50
    [cameron] another example of targeting ARIA in CSS
  106. zakim-robot
    18:58
    [jdan] @cameron @heydon touched upon this in https://www.smashingmagazine.com/2013/08/semantic-css-with-intelligent-selectors/
  107. zakim-robot
    19:04
    [fstorr] @cordelia: thanks for that info.
  108. zakim-robot
    19:09
    [jessebeach] @jnurthen, I'll keep it in mind as a possible trap. Thank you for noting it.
  109. zakim-robot
    19:10
    [jessebeach] Also, I'll talk with Matt King. Maybe getting out of a dialog requires a bit of advanced screen reader trickery that he just knows and applies intuitively at this point.
  110. zakim-robot
    19:39
    [cordelia] @cameron, we do a little bit of ARIA targetting in our CSS at dropbox, i’d love to do more
  111. zakim-robot
    19:42
    [cordelia] I also rather like what one of our designers added into Scooter (open-source CSS framework) for ensuring that things with a “label" class actually are <label>s: https://github.com/dropbox/scooter/search?utf8=%E2%9C%93&q=assert-selector
  112. MichielBijl
    20:00
    @cameron, want to do the test in a bit?
  113. MichielBijl
    20:01
    I must warn you: I'm already on my third whisky…
  114. garcialo
    20:10
    is this a cognitive impairment test?
  115. MichielBijl
    20:10
    Could be :P
  116. MichielBijl
    20:11
    It'll be a vision impairment test if I keep this up…
  117. MichielBijl @MichielBijl orders some Chartreuse
  118. powrsurg
    20:42
    Isn't this room available on slack and not just gitter?
  119. garcialo
    20:42
    correct
  120. jnurthen
    20:42
    yes - available on both
  121. deborahgu
    20:43
    and xmpp gateway :D
  122. jnurthen
    20:44
    can't give you the URL for slack though as it is blocked at my work.
  123. powrsurg
    20:44
    I just had to join slack for another thing so I guess I should only be in one
  124. deborahgu
    20:45
    I wish slack were blocked at my work. Have they successfully filled that a11y position, yet?
  125. garcialo
    20:45
    @powrsurg You'll want to go to http://web-a11y.herokuapp.com/ to get an invite
  126. deborahgu
    20:45
    Then I will find that person and camp out on their door with cookies and bribes.
  127. jnurthen
    20:46
    don't think so - some guy from slack mentioned they had an opening at the a11ybay meetup on monday
  128. deborahgu
    20:46
    sigh
  129. deborahgu
    20:46
    1. become everybody's favorite chat tool and dominate the market
  130. deborahgu
    20:46
    1. hire an accessibility person
  131. deborahgu
    20:46
    1. profit!
  132. garcialo
    20:46
    Maybe the chat should apply as a group and they can send us all portions of the salary.
  133. jnurthen
    20:46
    @deborahgu at least they are looking for someone..... it is a first step!
  134. MichielBijl
    20:46
    Ha
  135. deborahgu
    20:47
    true, @jnurthen!
  136. MichielBijl
    20:47
    They're not looking, they have a posting.
  137. MichielBijl
    20:47
    Nobody has a posting for six months in SV.
  138. MichielBijl
    20:47
    (I'm guessing)
  139. jnurthen
    20:48
    there are not many people with the experience they are looking for
  140. zakim-robot
    20:48
    [powrsurg] so seeing this room in slack and gitter at the same time is fun. In gitter @deborahgu's comment was an ordered list of 1. each time. In slack it went 1, 3, 2
  141. MichielBijl
    20:48
    on IRC it's 1,3,2 too
  142. jnurthen
    20:48
    which is better?
  143. jnurthen
    20:49
    1,1,1 or 1,3,2?
  144. MichielBijl
    20:49
    Well the order is wrong, so doesn't matter I guess.
  145. powrsurg
    20:49
    well, it worked for the joke ...
  146. jnurthen
    20:50
    LOL
  147. MichielBijl
    20:50
    @cameron I have other stuff to do, we can try tomorrow :)
  148. MichielBijl
    20:50
    Haha, yeah I guess
  149. jnurthen
    20:50
    drinking whisky isn't other stuff ;)
  150. powrsurg
    20:50
    hmm, slack sees whatever I type regardless of the room. Gitter only sees what I type here
  151. MichielBijl
    20:50
    You underestimate how much of my W3C stuff I do while drinking whisky :P
  152. jnurthen
    20:51
    same goes for my conference calls with india :)
  153. deborahgu
    20:51
    whoa! I just typed it as three separate lines.
  154. deborahgu
    20:51
    funny
  155. MichielBijl
    20:51
    that's probably why they think brits sound so weird ;)
  156. MichielBijl
    20:52
    I should probably get of the internet…
  157. jnurthen
    20:53
    the real world? why would you do that?
  158. MichielBijl
    20:53
    Because I need to clean the house
  159. MichielBijl
    20:53
    I could hire a cleaning service trough the internet…
  160. MichielBijl
    20:54
    Have you ever had Chartreuse?
  161. jnurthen
    20:54
    yeah..... not in my standard booze rotation though
  162. MichielBijl
    20:54
    It's really good.
  163. jnurthen
    20:55
    you can also pretend it is good for you
  164. MichielBijl
    20:55
    So who do I talk to to get paid to write standards :P
  165. MichielBijl
    20:55
    Nah, I've done that a while; it wasn't.
  166. jnurthen
    20:56
    does anyone pay people to write standards?
  167. MichielBijl
    20:56
    That's what I'm asking :)
  168. jnurthen
    20:57
    I think all of us have day jobs too
  169. MichielBijl
    20:57
    That's what I figured.
  170. jnurthen
    20:57
    luckily some of our companies give us some time to do this too......
  171. MichielBijl
    20:57
    Still, some time is better than no time at all.
  172. jnurthen
    20:57
    just have to find one of them
  173. jnurthen
    20:57
    or get W3C to hire you
  174. MichielBijl
    20:57
    To be fair thought, every spare minute at my job goes to ARIA stuff :P
  175. MichielBijl
    20:58
    But that is not because they say I can do that, just that they don't bother to put me on projects…
  176. MichielBijl
    20:58
    Doesn't the W3C mainly work with people sent from other companies?
  177. jnurthen
    20:59
    there are some staff
  178. MichielBijl
    20:59
    That's why I said mainly :P
  179. MichielBijl
    20:59
    But yeah, there is some staff.
  180. jnurthen
    20:59
    https://www.w3.org/People/
  181. jnurthen
    20:59
    you know quite a few of them
  182. MichielBijl
    21:00
    There aren't that many job openings: https://www.w3.org/Consortium/Recruitment/
  183. MichielBijl
    21:00
    True
  184. garcialo
    21:04
    seriously though, Standards work is tough; it really takes a toll on you
  185. garcialo
    21:04
    just look at @stevefaulkner he just pretends to be much older, but he's actually in his late 20s
  186. MichielBijl
    21:06
    Some say he's younger than me…
  187. garcialo
    21:07
    he's younger than us all
  188. powrsurg
    21:55
    Did that aria-errormessage thing ever get resolved?
  189. MichielBijl
    21:56
    as in?
  190. jnurthen
    22:15
    this is the current Editor draft - https://w3c.github.io/aria/aria/aria.html#aria-errormessage
  191. jnurthen
    22:15
    What was the question about aria-errormessage
  192. stevefaulkner
    23:28
    @powrsurg there is only 1 a11yskackers room, it's like slacks cattle class
  193. stevefaulkner
    23:29
    @MichielBijl w3c staff generally don't write specs, they are around to facilitate the suckers that do
  194. stevefaulkner
    23:30
    I sorta like this https://longtermlaziness.wordpress.com/2013/10/08/the-w3c-is-a-restaurant/
  195. stevefaulkner
    23:31
    THE W3C IS A RESTAURANT
  196. garcialo
    23:31
    I'll have a #3
  197. garcialo
    23:31
    hold the mayo
  198. stevefaulkner
    23:32
    Yesterday I was older then I am younger than that now - to paraphrase a byrds lyric
  199. stevefaulkner
    23:34
    Working from my 🚽 Again 😜
  200. garcialo
    23:34
    It's like how in space no one can hear you scream.
  201. garcialo
    23:35
    Except the chat is space, and the screaming is...well...poop.
  202. zakim-robot
    23:35
    [karlgroves] Never been in space. Wouldn’t know.
  203. stevefaulkner
    23:38
    Pleased that notes on using aria in HTML is included in this list https://twitter.com/sitepointdotcom/status/692428121660919808
  204. zakim-robot
    23:40
    [karlgroves] Steve Faulkner: Go to sleep, old man!
  205. garcialo
    23:43
    Oh geez, indeed. It's well after 7 p.m. London time, @stevefaulkner
  206. stevefaulkner
    23:44
    Normally in bed before 10 me self, but big day tomorrow...
  207. garcialo
    23:44
    well good luck to you then