Archive index

A11y Slackers Gitter Channel Archive 5th of October 2017

What fresh hell is THIS now? - Patrick Lauke
  1. zakim-robot
    @zakim-robot
    Oct 05 00:12
    [jhetrick] if there’s an element that will download a file (say, a zip or a pdf) — is that “triggering an action” and thus a button element, or “going somewhere” (to a file) and thus an a element?
  2. zakim-robot
    @zakim-robot
    Oct 05 00:34

    [svinkle] @jhetrick It’s a good question. I feel like it could go either way.

    With an a, the href value would serve as the path to the file for download. Problem is the focus remains on the a which sort of defeats conveying its semantic meaning.

    Using a button kind of makes sense, as you’re invoking an action that typically results with a dialog at the operating system level. That said, you’d need JS to go fetch the file on the client event, so that’s not ideal either.

    I’m learning towards a. For example, downloading a .zip archive from GitHub: 

    <a href="/[user]/[project]/archive/master.zip" class="btn btn-outline get-repo-btn " rel="nofollow" data-ga-click="Repository, download zip, location:repo overview">   Download ZIP </a>
  3. zakim-robot
    @zakim-robot
    Oct 05 00:41
    [jhetrick] thanks! I was leaning towards a as well, with “Download” somewhere in the link text
  4. [jhetrick] I’m thinking that in the case of a zip file, it’s only by convention that browsers don’t open them (the way that they might with a pdf). I could theoretically have a browser extension that previews the contents of a zip file, in which case the browser might navigate to the file preview rather than always downloading it
  5. [jhetrick] thus navigation, not action
  6. zakim-robot
    @zakim-robot
    Oct 05 00:54
    [ugi] @jhetrick afaik it’s possible to force any file to trigger a download dialog (with the Content-Disposition header), instead of being displayed in the browser window. Not sure if it will override a browser plugin, but worth a try if you have control over the server that hosts the zip file.
  7. zakim-robot
    @zakim-robot
    Oct 05 11:16

    [scottohara] Sounds like you've gotten this worked out, but my two cents here since I don't notice them mentioned above:

    The existence of the download attribute, for use on <a> elements, helped from allowing me to pontificate too hard on whether a vs button was the right element.

    also, you wouldn't want the downloading of a file to be tied to the need for JS. Browsers will navigate to and render file types that they can understand (when not using the download attribute, or if the attribute is not understood. So it is still properly serving as a link). And the browser will load download dialogs for those file types they can't render.

  8. zakim-robot
    @zakim-robot
    Oct 05 15:26
    [eric] chiming in that good UX is to specify the filetype and size somewhere in the link, so you’ll get a better idea of what program might pop open to view it, and how long it may take to pull the data down
  9. [eric] download attribute also lets you specify human-friendly names for the download, while maintaining machine-friendly names on the CDN
  10. [higley] Has anyone done much cross-device testing around around aria-activedescendant? Or any opinions on using it vs. moving actual focus? (In this case specifically for something like a combobox/listbox)
  11. [svinkle] <a href="path/to/file.pdf"> Download [unique identifier] <img src="images/pdf.png" alt=", opens PDF file"> </a>
  12. zakim-robot
    @zakim-robot
    Oct 05 15:31
    [svinkle] @higley I’ve only ever used aria-activedescendant when implementing an autocomplete widget. Even then, I wasn’t sure if it really help with any sort of context, but good to follow spec.
  13. zakim-robot
    @zakim-robot
    Oct 05 15:43
    [higley] I’ve had some issues with it reading the correct active option when navigating the list of options, especially using it in combination with aria-owns, like in a combobox (although maybe I’m doing something else wrong, and aria-owns is just incidental)
  14. [higley] in VO + Safari, specifically. NVDA seemed fine in the one use case I tried, and I haven’t tested JAWS
  15. zakim-robot
    @zakim-robot
    Oct 05 16:23
    [car] Welcome, @gibsonb!
  16. zakim-robot
    @zakim-robot
    Oct 05 16:50
    [svinkle] Hey @mwala! :wave:
  17. [mwala] Hi @svinkle!
  18. zakim-robot
    @zakim-robot
    Oct 05 18:13
    [starla175] Does anyone know if this site is no longer working? I am doing a mobile a11y presentation tomorrow and want to demonstrate how an accessible page sounds when it’s coded properly, and how an inaccessible page sounds, but even the “good” example is acting really janky. https://www.w3.org/WAI/demos/bad/after/home.html
  19. [starla175] I guess I should have said, it’s acting really janky on an iPad.
  20. [starla175] Nevermind… Problem solved. I was using Chrome.
  21. zakim-robot
    @zakim-robot
    Oct 05 18:21
    [jasonday] @starla175 - For my presentations, I have two different pages (simple login form) - one coded poorly for a11y and one coded well. I get a volunteer from the audience to use a screenreader going through each page and talk me through what they think is happening. It works really well impact-wise and because it is a focused feature, the user is not overwhelmed.
  22. zakim-robot
    @zakim-robot
    Oct 05 19:10
    [starla175] That’s a great idea @jasonday. I think I will do that. It’s hard comparing apples to oranges.
  23. zakim-robot
    @zakim-robot
    Oct 05 19:22
    [jamesn] I've noticed aria-activedescendent issues on iOS w/ VO and Safari. I need to get around to bugging it