1479239 - prefers-reduced-motion should be sensible with unsupported platforms / resistFingerprinting.

Status: RESOLVED FIXED

(Core :: CSS Parsing and Computation, enhancement)

Description

[Emilio Cobos Álvarez [:emilio]]

With the current early-returning behavior, the expression will unconditionally
evaluate to false, which is a bit of a footgun.

Comment 1

[Emilio Cobos Álvarez [:emilio]]

Attachment: Bug 1479239: prefers-reduced-motion should be sensible re. unsupported platforms / resistFingerprinting. r=hiro

With the current early-returning behavior, the expression will unconditionally
evaluate to false, which is a bit of a footgun.

Make sure to always return no-preference in unsupported platforms or when
resisting fingerprinting.

Comment 2

[Hiroyuki Ikezoe (:hiro)]

Oh thanks. The test case is really easy to understand what the problem is, Emilio is mentioning.

But hmm, it seems to me the expected background-color is 'red'.

Tom, what should we do when the resistFingerprinting pref is set?  We should do 'no-preference' or do as if we don't support prefers-reduced-motion at all?   It doen't expose any system settings, so it's not a big deal though.

Comment 3

[Emilio Cobos Álvarez [:emilio]]

How would the expected background color be red? That trivially exposes whether the resist-fingerprinting pref is enabled, which looks like an anti-goal to me :)

Comment 4

[Hiroyuki Ikezoe (:hiro)]

Oh right, indeed. It looks the pref is exposed actually. :)  Then the change makes totally sense to me.
Tom, sorry for the noise. :)

Comment 5

[Phabricator Automation]

Comment on attachment 8995763 [details]
Bug 1479239: prefers-reduced-motion should be sensible re. unsupported platforms / resistFingerprinting. r=hiro

Hiroyuki Ikezoe (:hiro) has approved the revision.

https://phabricator.services.mozilla.com/D2491

Comment 6

[Emilio Cobos Álvarez [:emilio]]

Note that pages right now can detect whether resist-fingerprinting is enabled via a similar offender (GetSystemMetric), with something like:

  let mediaList = matchMedia("(-moz-touch-enabled)");
  let parsed = mediaList.media != "not all";
  let resistFingerprintingEnabled =
    parsed && !mediaList.matches && !matchMedia("(-moz-touch-enabled: 0)").matches;

I'm fixing that in bug 1479240.

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/b1e7eb636989
prefers-reduced-motion should be sensible re. unsupported platforms / resistFingerprinting. r=hiro

Comment 8

[Natalia Csoregi [:nataliaCs]]

https://hg.mozilla.org/mozilla-central/rev/b1e7eb636989

Comment 9

[Brian Birtles (:birtles)]

This is probably a separate bug, but looking at this demo:

  https://webkit.org/blog-files/prefers-reduced-motion/prm.htm

The last example has:

  var motionQuery2 = matchMedia('(prefers-reduced-motion: no-preference)'); 
  function handleReduceMotionChanged2() {
      document.images['jaws'].src = motionQuery2.matches ? 'jaws.gif' : 'jaws.jpg';
  }
  handleReduceMotionChanged2(); // trigger this once on load to set up the initial value
  motionQuery.addListener(handleReduceMotionChanged2); // Note: https://webkit.org/b/168491

And the test case text has:

  Example 4: Only display animated version if prefers-reduced-motion is both supported *and* off.

  This example uses a more explicit match for (prefers-reduced-motion: no-preference) which excludes all browsers that don't yet  support the new feature.

For platforms where we don't support prefers-reduced-motion, should we still match prefers-reduced-motion: no-preference ?

On Linux I see the animated graphic although we don't meaningfully support prefers-reduced-motion there.

Comment 10

[Hiroyuki Ikezoe (:hiro)]

That's a trick of our ui.prefersReducedMotion preference.  On Linux we don't tie prefers-reduced-motion to the system setting yet (just landed on autoland though in bug 1478519), but we can set the preference value on Linux, and the preference value is 'no-preference' by default.

Comment 11

[Brian Birtles (:birtles)]

Oh I see. I didn't realize there was a pref for it. I thought it wasn't supported at all on Linux.