Archive index

A11y Slackers Gitter Channel Archive 22nd of December 2016

What fresh hell is THIS now? - Patrick Lauke
  1. zakim-robot
    @zakim-robot
    Dec 22 03:31
    [csantos] hello there, I have the following code:
    <div id=”container" tabindex=”-1”>Hello world</div>
    I’m doing $(’#container’).focus(), but NVDA-Firefox is only reading: _”region”_
  2. [csantos] what should I do?
  3. [caesar] Does it read "hello world" after a keypress (down arrow)?
  4. zakim-robot
    @zakim-robot
    Dec 22 03:36
    [csantos] no :|
  5. zakim-robot
    @zakim-robot
    Dec 22 03:41
    [csantos] mm I work in a Windows virtual machine from a Macbook, it’s so hard to press/send a right NVDA-key
  6. James Nurthen
    @jnurthen
    Dec 22 03:53
    Divs would not normally be mapped in the accessibility API. What is it? Why are you trying to focus to it? What does it do?
  7. zakim-robot
    @zakim-robot
    Dec 22 03:58
    [karlgroves] @csantos give that div a role.
  8. zakim-robot
    @zakim-robot
    Dec 22 04:11
    [csantos] the whole idea is: we have a link to expanded/collapsed that <div>, so, when that content is expanded, I’m forcing the focus to that div to be automatically read by NVDA
  9. Peter Krautzberger
    @pkra
    Dec 22 08:01
    Hiya slackers. Like @StommePoes I've been investigating SVG recently. I recorded some test results at https://codepen.io/pkra/pen/RodYOJ.
    I'm wondering if anybody has more data.
  10. Mallory
    @StommePoes
    Dec 22 11:07
    I pinged Leonie and Chaals about it, might corner them at CSUN about it
  11. I still haven't had time to finish my own tests, busy with getting drag and drop to work with as many screen reader/browser combos as I can...
  12. looks like I have a due date of 2 jan 2017
  13. zakim-robot
    @zakim-robot
    Dec 22 11:18
    [alastc] I've noticed @redux doing some SVG testing recently, not sure if he's on here much though.
  14. Mallory
    @StommePoes
    Dec 22 11:18
    Oh, I could ping him too I suppose
  15. Didn't know he was getting back to that
  16. zakim-robot
    @zakim-robot
    Dec 22 11:18
    [alastc] It was mostly complaints on twitter, don't get your hopes up!
  17. Mallory
    @StommePoes
    Dec 22 11:19
    hehe
  18. Well, the pointer events stuff was too... and it ended up in a lot of good info
  19. zakim-robot
    @zakim-robot
    Dec 22 11:37
    [dean] Question about “Skip links” - if a site uses good structural elements, does it make the need for skip links redundant? (or at least less important)
  20. [alastc] Something odd with VoiceOver, if you add tabindex (and use the tab key) you can get to the SVG & hear the titles. https://codepen.io/alastc/full/BQEdyw/
  21. Mallory
    @StommePoes
    Dec 22 11:41
    Skip links are important! for regular keyboard users
  22. forget screen reader users, regular keyboarders don't want the 1000 tabs of death to get to the main content
  23. esp on ecommerce sites
  24. zomg
  25. @alastc that may be because when you're tabbing you're in another mode (for whatever that means in VO which doesn't do modes the way the win sr's do)
  26. zakim-robot
    @zakim-robot
    Dec 22 11:43
    [alastc] I"m sure I've tested SVG in VO before and didn't have this issue of not being able to get to them. I wonder if it's a recent bug.
  27. [alastc] Or I'm getting confused with MathML, that's a distinct possibility.
  28. Mallory
    @StommePoes
    Dec 22 11:46
    @dean go to vanbeekdesign.nl and navigate to a page full of products. You'll immediately appreciate the skip links to "inhoud" (content)
  29. zakim-robot
    @zakim-robot
    Dec 22 11:48
    [dean] Thanks! You’re right, switched straight to screenreader user in my head, it’s a simple site with just 4 headings at the top, but yep it’s still repetitive.
  30. Mallory
    @StommePoes
    Dec 22 11:55
    If there's only a few tab stops then I wouldn't demand one
  31. but a "typical" site usually has a ginormous navigation
  32. I'm ok with 5-10 tab stops
  33. 10's pushing it tho
  34. or like Github's login page where you have to start shift-tabbing and you can't predict which button/link you'll land on first because they have sh*tty source-order...
  35. and set all their form fields in... dl's for some reason... without dt tags anyways... wtf github
  36. zakim-robot
    @zakim-robot
    Dec 22 11:58
    [dean] 6 tab stops.
  37. Mallory
    @StommePoes
    Dec 22 11:59
    ok, for 6 I wouldn't bother with skip links, esp if as you say the rest of the page is very well structured
  38. unless there's anything on the page most people would want to go to and it becomes a bazillion tab stops to that thing
  39. eventually somewhere something is a bazillion tab stops but for most things that's just expected
  40. zakim-robot
    @zakim-robot
    Dec 22 12:09
    [dean] thanks :)
  41. Peter Krautzberger
    @pkra
    Dec 22 15:05
    alastc no MathML involved in my samples; just SVG.
  42. thanks though for the data, alastc!
  43. zakim-robot
    @zakim-robot
    Dec 22 15:05
    [unknown] alastc no MathML involved in my samples; just SVG.
  44. zakim-robot
    @zakim-robot
    Dec 22 16:57
    [alastc] np, the MathML comments was just that I'd been testing some recently, some of which were displayed via SVG, but I can't remember the results off hand.
  45. Peter Krautzberger
    @pkra
    Dec 22 17:19
    Ah, that was probably using MathJax (which I'm involved in).
  46. zakim-robot
    @zakim-robot
    Dec 22 17:19
    [unknown] Ah, that was probably using MathJax (which I'm involved in).
  47. [alastc] @pkra ah, interesting. Our client (publisher) was looking for advice for their science journals. The options for equations in PDF seem very limited for a public audience, so we recommended focusing on the HTML and later ePub output, using MathML/MathJAX. Would you agree?
  48. Peter Krautzberger
    @pkra
    Dec 22 17:26
    Sounds about right but it's a tricky field.
  49. zakim-robot
    @zakim-robot
    Dec 22 17:26
    [unknown] Sounds about right but it's a tricky field.
  50. Peter Krautzberger
    @pkra
    Dec 22 17:26
    E.g., generating PDF from HTML works well with MathJax too, but e.g., SVG output is often poorly rendered by PDF viewers.
  51. zakim, stop repeating me...
  52. despite printing just fine.
  53. zakim-robot
    @zakim-robot
    Dec 22 17:27
    [unknown] E.g., generating PDF from HTML works well with MathJax too, but e.g., SVG output is often poorly rendered by PDF viewers.
  54. [unknown] despite printing just fine.
  55. Peter Krautzberger
    @pkra
    Dec 22 17:27
    I generally suggest pre-processing with mathjax-node these days. Unless you have dynamic content, there's really no need to use MathJax on the client.
  56. zakim-robot
    @zakim-robot
    Dec 22 17:28
    [alastc] From testing and checking other people's compatibility tables, only NVDA + the plugin was the only real option for PDF. HTML was better, and seemed to have a trajectory of improving, eventually.
  57. Peter Krautzberger
    @pkra
    Dec 22 17:29
    Yup. HTML is awesome. Just MathML is not.
  58. in particular has no trajectory of improving.
  59. zakim-robot
    @zakim-robot
    Dec 22 17:30
    [alastc] sorry, I meant support for it MathML in HTML appeared to be improving, a bit.
  60. Peter Krautzberger
    @pkra
    Dec 22 17:31
    Ah. ok, I disagree then but it doesn't matter much.
  61. zakim-robot
    @zakim-robot
    Dec 22 17:33
    [alastc] Probably doesn't make and different to what I would have recommended, which was to improve their alternatives and update their version of MathJAX, but in terms of user-agent support I was going from things like this http://msf.mathmlcloud.org/setups/results?utf8=%E2%9C%93&q%5Bplatform_version_platform_id_eq%5D=&q%5Bassistive_technology_version_assistive_technology_id_eq%5D=&q%5Bbrowser_reader_version_browser_reader_id_eq%5D=&q%5Bfile_format_id_eq%5D=3&q%5Brenderer_id_eq%5D=&q%5Boutputs_id_eq%5D=&q%5Bworkflow_status_id_eq%5D=&button=
  62. Peter Krautzberger
    @pkra
    Dec 22 17:40
    yeah, that's not a reliable source. Support in VO and JAWS is poor and unmaintained. Browser vendors have never implemented MathML -- it's all volunteer work. And finally the Working Group has been shut down this year, so the spec is effectively dead.
  63. So unless the last volunteer standing can magically convince Chrome and Edge to change their minds, there will never be proper support in browsers.
  64. and then you're still left with a fundamentally inaccessible standard since practically all MathML in the wild is Presentation MathML.
  65. Peter Krautzberger
    @pkra
    Dec 22 17:46
    Essentially, all existing MathML accessibility tools (including our sister project speech-rule-engine) have to rely on heuristics to guess semantics where none exist. It's not pretty and might as well be slapped onto HTML or SVG; there's really no need for MathML.
  66. Oh dear, there I rant again. Better call it a day ;-)
  67. zakim-robot
    @zakim-robot
    Dec 22 22:16
    [caesar] I vaguely remember seeing something on the WCAG notes about skip links being required if there are 4-6 repeated links on a page...? Or did I imagine that?