Archive index

A11y Slackers Gitter Channel Archive 20th of December 2016

What fresh hell is THIS now? - Patrick Lauke
  1. James Nurthen
    @jnurthen
    Dec 20 00:03
    if the parent is a menu then the action "buttons" must be menuitems
  2. there would be no buttons in the accessible tree in this case
  3. Mallory
    @StommePoes
    Dec 20 09:56
    @powrsurg I've had it the other way around: having something running like chromecast has crashed my ZT
  4. Peter Krautzberger
    @pkra
    Dec 20 10:41
    Morning slackers. I just ran into a case where VoiceOver changed the order of what it read to match the visual order rather than something more like the DOM order (the code is messy, absolutely positioned span soup). Is this a well known thing?
  5. Mallory
    @StommePoes
    Dec 20 10:49
    how were they moved around? flexbox? floats?
  6. in Safari?
  7. mobile or no? I thought on mobile it really did follow screen position more than on desktop
  8. Peter Krautzberger
    @pkra
    Dec 20 11:07
    hey @StommePoes just positioning. I was testing Safari 9 on deskktop.
  9. e.g. 3. on https://codepen.io/pkra/pen/VmZLgZ
  10. those sums from k=1 to n
  11. The dom structure would indicate n => sum => k=1 (but there are a bunch of lables mucking about)
  12. but Safari reads it differently (the "correct" way: sum => k=1 => n)
  13. I'll try to build a minimal example.
  14. Mallory
    @StommePoes
    Dec 20 11:26
    is safari the only one reading it "correctly"?
  15. and yeah, "k" should be ignored if you've got an aria-label and the AT/browser decide that the aria-label should be doing labeling of the thing-- if they don't, you'd only hear k and not the aria-label.
  16. so, did you have to manually put in those aria-labels or did mathjax do that?
  17. does anything get read out with no aria-labels?
  18. Peter Krautzberger
    @pkra
    Dec 20 11:31
    Safari was the only one so far. NVDA ignores most of it, JAWS decided to crash while re-installing.
  19. Mallory
    @StommePoes
    Dec 20 11:31
    and only NVDA uses mathplayer right?
  20. or was it just that NVDA had a plugin-version of it? I remember testing their test page before and after and both read out fine, but something I was testing said mathplayer was required for students
  21. (for that content)
  22. Peter Krautzberger
    @pkra
    Dec 20 11:34
    mathplayer does not come into play.
  23. there's no MathML anywhere.
  24. :)
  25. this is just about HTML (and also SVG but that's a different story)
  26. Mallory
    @StommePoes
    Dec 20 11:35
    No was just asking about mathplayer in general
  27. turns out newer versions of Office have it (or a lite version) but older versions don't and people need to know to go download it
  28. for powerpoints
  29. But
  30. I thought that <math> tags etc... was mathml
  31. Peter Krautzberger
    @pkra
    Dec 20 11:37
    ah, ok. NVDA uses it (and ChromeVox of course).
  32. the input is mathml but the output is not.
  33. Anyway. My sample is clearly causing more confusion -- I'll make a minimal one.
  34. sorry about that! and thank you!
  35. Mallory
    @StommePoes
    Dec 20 11:38
    lawlz
  36. I didn't know about chromevox
  37. I really ought to learn more about them since schools seem to be turning to them massively
  38. chromebooks
  39. Peter Krautzberger
    @pkra
    Dec 20 11:49
    right.
  40. a bit annoying that the chromevox2 beta is currently chromOS-only.
  41. Mallory
    @StommePoes
    Dec 20 11:50
    I thought that's what it was built for, chromeOS
  42. Peter Krautzberger
    @pkra
    Dec 20 11:50
    it's a completely different beast than v1.
  43. Mallory
    @StommePoes
    Dec 20 11:50
    like, back when Google thought "we don't need to support AT, we'll just build our own out of JS"
  44. Peter Krautzberger
    @pkra
    Dec 20 11:50
    Yes, but it also exists as a Chrome extension.
  45. Mallory
    @StommePoes
    Dec 20 11:50
    Oh, didn't know. Does v2 act completely different? still built in JS?
  46. Peter Krautzberger
    @pkra
    Dec 20 11:50
    so you can test ChromeVox v1 in any Chrome.
  47. Mallory
    @StommePoes
    Dec 20 11:51
    I didn't know... wonder if even on linux? if it's chromium does it work?
  48. most extensions do work on chromium
  49. Peter Krautzberger
    @pkra
    Dec 20 11:51
    It's still built in JS but AFAIK it no longer lives in the DOM itself, just accesses the a11y tree .
  50. yes, it works on linux.
  51. it's just JS after all and ChromeOS is Ubuntu-based.
  52. Mallory
    @StommePoes
    Dec 20 11:53
    oh, well it's good that it just accesses the tree
  53. it's where everything's going at this point anyway
  54. like Edge
  55. They're calling it a sec issue, which maybe it is. Jamie said with FF he only accesses the DOM for MathML actually.
  56. I filed a FF bug, turns out it might really be a real bug.
  57. for reading out SVG
  58. You might run into it... NVDA will read out SVGs in FF but JAWS and Orca won't (or my version of Orca won't, Joanie said it worked for her but she was running latest)
  59. Peter Krautzberger
    @pkra
    Dec 20 11:55
    Right.
  60. Saw your bug report yesterday.
  61. but NVDA just ignores all my fancy labels :-(
  62. on HTML.
  63. I'll have to write to Jamie eventually ;-)
  64. Mallory
    @StommePoes
    Dec 20 11:57
    heh
  65. Or Derek Riemer
  66. Marco commented today
  67. on my bug. I wasn't entirely sure if it was a bug, but despite dying of moz-flu he advised me to file it anyway
  68. Because I know JAWS does some crazy stuff sometimes, and I have no idea if it DOM-dumpster-dives in FF or not for SVG.
  69. Hm, I'm curious if your aria-labels are not being interpreted by the browser as labels... math tags might not be considered labellable by the browser
  70. I could look in the Dom Inspector tool but I was warned it's not fully reliable nowadays.
  71. Peter Krautzberger
    @pkra
    Dec 20 12:04
    don't worry about the math tags.
  72. MathJax does pick them up and puts them in the right places.
  73. the labels on the math tags, that is.
  74. again, sorry for giving such a confusing example.
  75. Mallory
    @StommePoes
    Dec 20 12:06
    lawlz
  76. it's onl confusing because I know so little about it
  77. Peter Krautzberger
    @pkra
    Dec 20 12:08
    :)
  78. nobody knows anything about this.
  79. Mallory
    @StommePoes
    Dec 20 12:11
    Somebody knows something, just not me!
  80. Peter Krautzberger
    @pkra
    Dec 20 12:12
    Mpf. I can't reproduce a minimal example...
  81. Mallory
    @StommePoes
    Dec 20 12:13
    Edgar Lozano knows something... he worked with Sam Dooley on that nemeth mathml thing
  82. Peter Krautzberger
    @pkra
    Dec 20 12:17
    I have not yet met Edgar, actually.
  83. Mallory
    @StommePoes
    Dec 20 12:20
    found him on twitter, asked if he's CSUNning next year. Might not tho
  84. he and Su Park were basically the exhibitors of the mathML-to-braille/braille-to-mathML translator.
  85. Peter Krautzberger
    @pkra
    Dec 20 12:22
    Cool.
  86. thanks for sharing.
  87. Yikes, my mistake. My Safari setup was doing something different from my other browsers which affected the DOM structure.
  88. Sorry for wasting your time @StommePoes :bow:
  89. Mallory
    @StommePoes
    Dec 20 12:31
    lawlz
  90. But, what was Safari doing?
  91. Is it a known thing?
  92. like FF being different with flexbox?
  93. (I could easily make the same mistake, so it's good to know what can happen)
  94. Peter Krautzberger
    @pkra
    Dec 20 13:08
    No, Safari was doing everything as I would expect.
  95. I was simply looking at two different DOM structures thinking they were the same.