Archive index

A11y Slackers Gitter Channel Archive 17th of December 2015

What fresh hell is THIS now? - Patrick Lauke
  1. MichielBijl
    08:44
    Where can I file bugs against Narrator/MS software?
  2. pkra
    09:54
    I was wondering what people's experience is with identical AT+browser versions but different OS versions. E.g., in what I'm testing right now, NVDA+FF seems to behave quite the same from XP to Win 10, so for a quick test it seems looking at one would be sufficient. But perhaps I'm just being lucky?
  3. stevefaulkner
    09:57
    @pkra OS version does not generally make a difference on windows when using same AT/browser combo
  4. stevefaulkner
    09:57
    and AT/browser versions
  5. MichielBijl
    09:57
    The browser has to expose the roles etc to the accessibility tree right? As long as the platform supports the feature, it should work, right?
  6. stevefaulkner
    09:58
    You will find a difference on iOS/OSX as the browser/AT versions are linked to the OS version
  7. pkra
    09:58
    @stevefaulkner thanks!
  8. pkra
    09:59
    @MichielBijl that's what I would think but I'm still such a n00b...
  9. MichielBijl
    09:59
    I still find it weird that VO differs between iOS and OS X.
  10. pkra
    09:59
    @stevefaulkner re iOS/OSX right.
  11. MichielBijl
    09:59
    VO on iOS seams to support more stuff :/
  12. pkra
    10:03
    @MichielBijl that is sad...
  13. MichielBijl
    10:04
    That is what I get from various tests, but not main stream usage.
  14. MichielBijl
    10:04
    You can see here: https://rawgit.com/MichielBijl/language-switch/master/tests/index.html
  15. MichielBijl
    10:05
    And here: https://rawgit.com/MichielBijl/font-awesome/master/tests/index.html
  16. MichielBijl
    10:05
    VO on iOS generally does a better job in those scenario's.
  17. pkra
    10:06
    Nice tests though. :)
  18. MichielBijl
    10:08
    Thank you, not possible without the help of the community =)
  19. pkra
    11:00
    I'm running into a curious problem with JAWS and Chrome. Instead of an aria-label, its reads the page title tag (from the head). Has anyone run into this?
  20. stevefaulkner
    11:09
    @pkra note that JAWS does not officially support chrome, but mostly works OK, in order to know where issue lies, need test file/example where it happens also JAWS/chrome version
  21. pkra
    11:11
    @stevefaulkner ah, thanks! I did not know that that's not officially supported. Just in the middle of testing. Should be merged later and pushed to gh-pages. Will post update.
  22. stevefaulkner
    11:12
    :+1:
  23. MichielBijl
    11:13
    @pkra what sort of test are you working on?
  24. pkra
    11:14
    A change to the math tool I linked to a while back.
  25. pkra
    11:14
    https://github.com/mathjax/MathJax-RespEq/pulls
  26. pkra
    11:15
    or this one to be precise mathjax/MathJax-RespEq#52
  27. MichielBijl
    11:15
    Cool!
  28. pkra
    11:35
    yeah, it's been going well. Except for a regression with WindowEyes on Chrome.
  29. pkra
    11:35
    oh well.
  30. MichielBijl
    11:37
    Does VO support mathy stuff?
  31. pkra
    11:46
    @MichielBijl VO has some (basic) voicing for MathML; it's far from perfect and bugs are not getting fixed :(. Though Safari has some nice exploration features on iOS for visual users (popup zoom lets you explore subexpressions) and in parts is better than what's missing in terms of visual rendering support in WebKit...
  32. pkra
    11:47
    so, kind of a mess.
  33. pkra
    11:47
    but very elementary math should be ok.
  34. MichielBijl
    11:50
    See, iOS is more advanced than OS X :P
  35. pkra
    11:53
    Hehe. OSX is much the same (and you have other means of zooming ;-) )
  36. pkra
    11:55
    Though of course VO on OSX does not support other browsers.
  37. pkra
    11:55
    for MathML.
  38. stevefaulkner
    12:16
    Nascent work on "What should an AT convey to users?" http://thepaciellogroup.github.io/AT-browser-tests/AT.html inital test files for <a href> and <a> in response to James teh's article http://blog.jantrid.net/2015/12/woe-aria-surprisingly-but-ridiculously.html PR's and bugs welcome!
  39. MichielBijl
    12:20
    @stevefaulkner is there a better way to record VO audio than through a microphone?
  40. MichielBijl
    12:20
    I'm afraid recordings will have too much background noise.
  41. stevefaulkner
    12:22
    i believe so, will ask someone who knows , i am a PC
  42. stevefaulkner
    12:33
    @MichielBijl seems like soundflower is used https://code.google.com/p/soundflower/
  43. MichielBijl
    12:33
    Thanks, I'll check that out.
  44. MichielBijl
    12:36
    Hope it still works (last updated three years ago).
  45. stevefaulkner
    12:40
    @MichielBijl also this via Hans VoiceOver (El Capitan): Repeat, copy, or save the last spoken phrase https://support.apple.com/kb/PH22525?viewlocale=en_US&locale=en_US
  46. MichielBijl
    12:41
    😱😱
  47. MichielBijl
    12:41
    Awesome!
  48. MichielBijl
    12:42
    FYI: also works on Yosemite.
  49. stevefaulkner
    12:48
    "If you need more than the last command I would use audacity and sound flower" - @hanshillen
  50. MichielBijl
    12:49
    :+1:
  51. MichielBijl
    12:51
    Cool thing about the VO way is that it also saves a txt with a string of what is spoken.
  52. garcialo
    12:53
    Audacity is available on Mac OS?
  53. garcialo
    12:54
    So it does. =)
  54. MichielBijl
    12:54
    http://www.audacityteam.org/
  55. MichielBijl
    12:54
    Yep
  56. stevefaulkner
    12:54
    http://audacityteam.org/download/mac
  57. stevefaulkner
    12:54
    through the magic of google it is found
  58. garcialo
    13:00
    What's that? I love magic.
  59. MichielBijl
    13:00
    Try DuckDuckGo, they only do magic in private though.
  60. MichielBijl
    13:03
    Looking for a good domain name is like going on a spiritual journey through different worlds and finding yourself lost in spaces filled with words that have vague meanings.
  61. garcialo
    13:05
    Just go with a nonstandard tld
  62. MichielBijl
    13:09
    alleycat.cafe is still free :')
  63. zakim-robot
    13:18

    [karlgroves] > Looking for a good domain name is like going on a spiritual journey

    “Back in the day” domain names were so easy to find.

    Tenon.io actually got its name after a month-long automated search.

    Here’s what I did:

    1. Get a plain text list of all words in the English dictionary

    2. Purge all words > 7 characters long, because we wanted something easy to type in CLI.

    3. Write shell script to go through the dictionary list one word at a time, append ‘.io’ to it (because all the kewl kids have .io names) and then do a WHOIS on that domain name

    4. If the domain name is available, append it to a file of available names.

    5. Go through the available names one by one and get rid of all gibberish like ‘abtfrw.io’ (ok that’s not a real word, but you know what I mean)

    6. Go through all the remaining names and find one you like. This was about 7000 names.

    eventually we narrowed it down to about 200 words that were at least actual, sensible, non-offensive words.

    Funny thing is that although I do woodworking, Asa was the one who said “Oooooh Tenon! I like that!” and thus we had our product name and domain name.

    FWIW, at least two years ago you pretty much had your pick of almost any offensive word + ‘.io’. We had a good time chuckling like junior high kids at all the offensive name possibilities.

  64. MichielBijl
    13:21
    Haha
  65. MichielBijl
    13:21
    That is a good idea actually. Alas, I on't know the first thing about scripting :P
  66. MichielBijl
    13:22
    Well, I could probably figure it out in the end.
  67. zakim-robot
    13:22
    [karlgroves] Do you know any backend languages? ASP.net? Node? PHP?
  68. MichielBijl
    13:23
    No, just the front-end stuff. I can get by in JS. So could write something that would interact with an WHOIS API probably.
  69. zakim-robot
    13:24
    [karlgroves] If I was smart I’d have saved that stuff, but I deleted it when I was done. :disappointed:
  70. MichielBijl
    13:24
    :(
  71. zakim-robot
    13:30

    [karlgroves] Oooooh. This looks easy to play with
    https://www.npmjs.com/package/node-whois

    If I get time, I’ll try to find that dictionary dump.

    Basically in Node you’d use FS to read the file, then run the whois, then write the file.

    The reason why it took a month, BTW, is because I didn’t want to have more than one whois request per minute, otherwise I’d get blocked for abuse

  72. MichielBijl
    13:32
    bookmarks
  73. MichielBijl
    13:38
    Looks interesting: https://whoiszz.com/api
  74. zakim-robot
    13:41
    [karlgroves] Hot damn! 10,000 requests per hour!
  75. MichielBijl
    13:41
    =D
  76. MichielBijl
    13:43
    Example: https://whoiszz.com/api/v1/whois/michielbijl.nl
  77. MichielBijl
    13:43
    Gives a 400 response if not found, like so: https://whoiszz.com/api/v1/whois/michielbijl.dk
  78. zakim-robot
    13:44
    [karlgroves] You’d have to do some string foo on the response, but that would work great
  79. MichielBijl
    13:45
    Only write out the 400 responses right? Those are all the free ones. Just ignore the json stuff :P
  80. zakim-robot
    13:45
    [karlgroves] Do they actually return proper status codes?
  81. MichielBijl
    13:46
    Not sure what a 400 response is.
  82. MichielBijl
    13:47
    If it doesn't exist, it doesn't return any json as far as I can see.
  83. zakim-robot
    13:48

    [karlgroves] Yeah, they do

    https://whoiszz.com/api/v1/whois/fkjhfghgg.io

    Returns HTTP 200 status

  84. zakim-robot
    13:48
    [karlgroves] and JSON
  85. zakim-robot
    13:49

    [karlgroves] So what you’d have to do to determine whether the name is available is do a string search for

    "Domain [your requested domain] is available for purchase"

  86. MichielBijl
    13:49
    That is weird, it does on refresh :/
  87. MichielBijl
    13:50
    It's very inconsistent for me.
  88. MichielBijl
    13:51
    Ah, depends on the domain name
  89. MichielBijl
    13:51
    sorry
  90. MichielBijl
    13:51
    TLD
  91. MichielBijl
    13:51
    Seems it doesn't do DK names
  92. MichielBijl
    13:52
    Hmm, it does, but only if they exist.
  93. MichielBijl
    13:52
    Like I said, inconsistent :P
  94. zakim-robot
    13:53
    [karlgroves] heh
  95. cptvitamin
    14:42
    @marcysutton yes, I did mix text content into all tests. I’ll clean up the test file and post it publically so others can test.
  96. pkra
    14:52
    @stevefaulkner I don't have a minimal sample but http://mathjax.github.io/MathJax-RespEq/Semantics-Lab/walker/ select one of the examples and the resulting equation will have role=application with an aria-label="math" -- but JAWS will read the title of the page. I have to work on something else now but can try to create something more minimal.
  97. stevefaulkner
    15:06
    @pkra will take a look
  98. pkra
    15:20
    @stevefaulkner thanks so much!
  99. zakim-robot
    15:30
    [johnbhartley] @MichielBijl not sure if you're still looking, but I use iShowU HD to capture my screen as well as the audio from VO when doing screen captures
  100. MichielBijl
    15:31
    Thank you @johnbhartley, but iShowU is paid IIRC. QuickTime does the same for free, but cannot capture internal audio as far as I know.
  101. MichielBijl
    15:32
    I use the microphone to capture it, but that also has all background noise.
  102. powrsurg
    15:32
    Has anyone seen today's 24ways article on accessibility? https://24ways.org/2015/the-accessibility-mindset/
  103. powrsurg
    15:33
    When talking about how to do buttons they mentioned: One way to tackle this is using the old-fashioned img element with an appropriate alt attribute, but surprisingly not every browser displays the alternative text visually when the image doesn’t load.
  104. powrsurg
    15:33
    what browsers don't show the alt text if the image doesn't load?
  105. MichielBijl
    15:35
    Safari doesn't if it doesn't fit in the space reserved for the image IIRC.
  106. MichielBijl
    15:35
    Maybe @yatil has done research on this?
  107. powrsurg
    15:35
    Lovely. I feel like they're pushing for icon fonts because of that reason ...
  108. MichielBijl
    15:36
    Those don't show alt text ever :P
  109. MichielBijl
    15:36
    And that's another reason buttons should have labels.
  110. MichielBijl
    15:36
    FYI: I still need to read that article.
  111. powrsurg
    15:37
    it does talk about adding labels
  112. powrsurg
    15:37
    basically all it gets into is adding labels to forms and talking about buttons
  113. MichielBijl
    15:39
    I'll read it after Marcy's interview =)
  114. zakim-robot
    15:41
    [johnbhartley] I was under the impression that title isn't read consistently across AT
  115. LjWatson
    15:43
    @johnbhartley the title attrib isn't supported consistently by screen readers, unless it's on a form field. It's also unsupported on touch devices.
  116. powrsurg
    15:43
    if title and aria-label are on is it double read?
  117. LjWatson
    15:44
    @powrsurg No. The accessible name computation means that the ARIA trumps the title attrib.
  118. zakim-robot
    15:45
    [johnbhartley] for more on title https://www.paciellogroup.com/blog/2012/01/html5-accessibility-chops-title-attribute-use-and-abuse/ h/t @stevef
  119. MichielBijl
    15:45
    http://www.w3.org/TR/2015/WD-accname-aam-1.1-20150714/#mapping_additional_nd_te
  120. LjWatson
    15:46
    The acc name and desc computation is here http://www.w3.org/TR/accname-aam-1.1/
  121. LjWatson
    15:46
    You may need a large drink after you read it though :(
  122. MichielBijl
    15:46
    @LjWatson sorry for my tweet, I was a bit to quick on the draw.
  123. MichielBijl
    15:47
    And I need more coffee before I post more links…
  124. LjWatson
    15:48
    @MichielBijl Oh. Your RT? It's ok, when I edited the original tweet, it automatically updates any RTs.
  125. MichielBijl
    15:50
    @LjWatson no I commented on the first tweet that the link was missing, but have deleted that tweet.
  126. LjWatson
    15:50
    @MichielBijl Aha! Missed that completely :)
  127. MichielBijl
    15:51
    How do you even edit a tweet :o
  128. LjWatson
    15:51
    In the app I use there is a menu/key command for editing a tweet.
  129. LjWatson
    15:51
    Guess it's an API feature of Twitter that some apps use and others don't?
  130. MichielBijl
    15:53
    Cool. I would think it just deletes the old and posts new tweet.
  131. MichielBijl
    15:54
    That's what I do anyway.
  132. LjWatson
    15:54
    Possibly, but I don't think so. It seems to propagate the change through any RTs etc. too.
  133. MarcoZehe
    15:57
    Hi folks! I would never ever have expected my Web Accessibility Basics post to go through the roof like that. Way over 5,000 views since I published it on Monday. And I thought I was telling people old news with this stuff!
  134. pkra
    15:57
    @MichielBijl I never got around trying this with quicktime https://github.com/mattingalls/Soundflower/releases
  135. MichielBijl
    15:59
    @MarcoZehe I have yet to read it, but from what I glanced it lists a lot of information on a page, people like that, makes for a good resource :P
  136. MichielBijl
    15:59
    @pkra ah, that is an updated version from what @stevefaulkner posted earlier. I'll try that too.
  137. stevefaulkner
    16:01
    @pkra FYI the textarea needs associated label <label for="input">Enter TeX code here:</label>
  138. stevefaulkner
    16:20
    @pkra have had a quick look and sort of worked out what is causing issue in chrome, don't have a lot of time right now, will get back to you with some suggestions
  139. zakim-robot
    16:24
    [yatil] @michiel: Safari is not showing the alternative text at all if the image is too small in width, IE afaik truncates the alternative text to the width/height of the image. So they are bold mildly useless when the image does not load. Firefox just replaces the image with a text node, which is the best version and picks also up the text styling without weird borders and spacing. http://codepen.io/yatil/pen/rxeemm
  140. zakim-robot
    16:25
    [yatil] needs to include this in my talk “36 ways to embed an image to an HTML page… And you’re still doing it wrong.”
  141. MichielBijl
    16:26
    If Firefox' UI wasn't so crappy I'd use it over Safari.
  142. zakim-robot
    16:27
    [yatil] I think the UI of the developer edition is pretty decent, but that is a tangent :wink:
  143. MichielBijl
    16:28
    The Developer version is definitely better.
  144. MichielBijl
    16:33
    And thank you for you elaboration @yatil =)
  145. MichielBijl
    16:33
    Am reading your article now.
  146. jasonday
    16:57
    Hi all.
  147. MichielBijl @MichielBijl waves at @jasonday
  148. garcialo
    16:57
    howdy
  149. jasonday
    17:00
    Question for the group - any recommendations on self hosted automated testing (includes references to objects, unlike Pa11y)? We are looking at commercial vendors (DeQue, BOIA, etc.), but I'd like to explore an internal tool (if feasible). For background, we are a large retailer - lots of visits, lots of complicated code, lots of legacy code as well
  150. MichielBijl
    17:03
    Tenon.io?
  151. MichielBijl
    17:04
    I believe they're looking at self hosting.
  152. garcialo
    17:04
    If you don't want to pay money, you could probably set something up internally using the Google Accessibility Dev Tools or aXe
  153. garcialo
    17:05
    if you want to pay money and set something up on your own, there's Tenon.io/WAVE API
  154. garcialo
    17:05
    if you want commercial, there's all that other stuff (WorldSpace, Compliance Sheriff, Tenon, etc.)
  155. zakim-robot
    17:11
    [karlgroves] @Jasonday you should come up with a list of requirements first for what you’d like such a tool to do and then do a basic “build or buy” analysis. When/ if the time comes to explore commercial options, you can use those same requirements to determine which product most closely suits your needs
  156. zakim-robot
    17:12
    [karlgroves] Follow that up with an actual trial of each product,
  157. garcialo
    17:13
    yeah
  158. garcialo
    17:13
    agreed =)
  159. zakim-robot
    17:15

    [karlgroves] FWIW the single most important important product feature is whether the product is capable of testing the rendered DOM of the page. That basically means it either exists in a browser or uses a headless browser.

    http://www.karlgroves.com/2013/09/06/web-accessibility-testing-tools-who-tests-the-dom/

    For the most part that limits you to Tenon, WAVE, Google Dev Tools, aXe, SortSite, Opquast, and Worldspace. (I have no data on AMP or Compliance Sheriff)

  160. deborahgu
    17:43
    For me one of the most important aspects of the product is can it run on a site behind auth / paywall.
  161. jasonday
    18:08
    Thanks everyone. I've done the bulk of requirements already against commercial products. @deborahgu DeQue is the only commercial product I found that had scripting capabilities , which allowed for getting behind auth areas (login, checkout, etc.), triggering behavior on pages (dynamic content, specific field values, etc.), and support for ARIA tags.
  162. deborahgu
    18:10
    @jasonday, so deque can run in full behind auth? right now I'm using Google, and it's fine, but more choices are always good.
  163. jasonday
    18:16
    @deborahgu - I haven't tested it hands on, but from the guided demo and the response from questions, there is the ability to script. You can trigger usernames/passowrds of testing accounts to have the test run those areas of your site (fully automated)
  164. deborahgu
    18:18
    cool, thanks
  165. MichielBijl @MichielBijl just wasted another hour discussing longdesc
  166. MichielBijl
    18:38
    When will it end?!
  167. MichielBijl
    18:40
    On a positive note, we sort of landed on http://www.w3.org/TR/wai-aria-practices/#Descriptions_external
  168. MichielBijl
    18:40
    Which is a couple years old…
  169. pkra
    19:32
    @stevefaulkner thanks!
  170. pkra
    21:24
    Here's something we cooked up this year, taking a page out of the a11y book to enhance the visual end https://www.mathjax.org/responsive-equations/
  171. zakim-robot
    21:52
    [ldavis] I have an audio button that when pressed plays some audio (click handler and audio.play()). When using JAWS I navigate to the button and it reads the aria-label, great. I then click space, which registers the click and plays the audio, however this according to JAWS appears to also bring the button into focus which at the same time as the audio playing reads the aria-label again and "press spacebar to activate". I cannot for the life of me figure out a meaningful way to play the audio and not have it reading back to me at the same time. Using tab seems to be the only way, but that doesn't solve the JAWS keyboard navigation to the button.
  172. zakim-robot
    21:59
    [ldavis] There doesn't seem to be a recognizable state difference between the first and second click of the button, while using JAWS, when navigating to buttons using their hotkeys that I can use to differentiate either...
  173. MichielBijl
    22:00
    In VO you can do modifier keys + spacebar to activate. Is there something similar in JAWS? Maybe that'll activated in a different way?
  174. zakim-robot
    22:05
    [ldavis] It seems just spacebar "activates" the button, then reads that spacebar can be used to trigger it, however that initial spacebar press also fires any click events hanging off it.
  175. zakim-robot
    22:07
    [ldavis] and they seem to look identical... And that button is the active element even though the UI doesn't necessarily reflect a state change. I may have to reach out to freedom scientific on this one too...
  176. zakim-robot
    22:26
    [devonpersing] ldavis, might be a good opportunity to test with users to see what they’d expect that kind of control to read as in those circumstances.
  177. zakim-robot
    22:37
    [astevens] @ldavis: maybe this is a silly suggestion, but toggle aria-hidden=“true” when audio starts and false when finished/paused/etc.?
  178. zakim-robot
    22:38
    [ldavis] I did just notice that there is no keydown event for the JAWS initial spacebar selection and should be able to take advantage of that.
  179. zakim-robot
    22:38
    [ldavis] Interesting astevens will give that a try too
  180. stevefaulkner
    22:51
    @ldavis test file for issue? Also browser and jaws version?
  181. zakim-robot
    22:53
    [ldavis] I'll have to put together a simple example of this (currently working with a private app) - Basically a button that plays an html5 audio. Otherwise Chrome 47.0.2526.106 and JAWS J17.0.1214enu-x64
  182. MichielBijl
    22:55
    @pkra note that JAWS does not officially support chrome, but mostly works OK, in order to know where issue lies, need test file/example where it happens also JAWS/chrome version
  183. MichielBijl
    22:55
    https://gitter.im/w3c/a11ySlackers?at=567297fa3e71e0707f9bb6bb
  184. zakim-robot
    22:58
    [ldavis] Interesting was not aware of Chrome support limitation. I'll put together a simple example tomorrow and post up