Re: [mediaqueries] light-level

> On Feb 25, 2016, at 05:40, fantasai <fantasai.lists@inkedblade.net> wrote:
> 
> On 02/09/2016 11:24 PM, Tab Atkins Jr. wrote:
>> On Tue, Feb 9, 2016 at 7:39 PM, Florian Rivoal <florian@rivoal.net> wrote:
>>>> On Feb 10, 2016, at 10:14, Tab Atkins Jr. <jackalmage@gmail.com> wrote:
>>>> We don't generally care about groups which need *less* a11y help, so
>>>> "reducing contrast" isn't really a use-case in the first place. ^_^
>>> 
>>> Actually it is. Not for people with low vision, but for people with
>>> certain forms of dyslexia, high contrast can make the text appear
>>> to be shining/sparkling/dancing, or just generally compound the
>>> difficulty of differentiating certain letter shapes (p vs q) and
>>> make things hard to read.
>>> 
>>>> On the other hand, increasing contrast for "light-level: washed" is a
>>>> good idea, *and* it can help with a11y that wants high-contrast.  (It
>>>> also generally means going with dark-on-light, which is better for
>>>> low-sighted users too.)
>>> 
>>> Luckily, this works in the other direction as well. The type of corrections
>>> that people tend to apply in response to "light-level: dim" also correlate
>>> well with the type of adjustment people with dyslexia favor:
>>> * reducing contrast
>>> * reducing blue light and going for a sepia / warm colors
>>> * light-on-dark
>> 
>> Good points!
> 
> My point isn't that there isn't overlap here, just that there
> are many cases where the response to light-level might not be
> something you want for a specific a11y concern. For example
> light-level responses can include
>  * increasing/decreasing contrast
>  * swapping light-on-dark vs. dark-on-light themes
> However, a11y preferences are generally one or the other.
> 
> So I would rather we
> 
>  a) Dealt with light-level and a11y issues around contrast
>     at the same time, but not in the same query

I am not sure I understand what you mean.

>  b) Instead advocated the use of the relevant a11y query
>     together with light-level -- e.g. making sure all examples
>     look like
>     @media (light-level: dim) or (pref-contrast: low) {
>        /* low contrast styles */
>     }
>     @media (light-level: dim) or (pref-foreground: light) {
>        /* light text on dark background */
>     }

I think that's the wrong way to go. Most authors are not
going to pay attention to a11y queries, so regardless of
what we put in the spec, they'd typically just write

@media (light-level: dim) { ... }

You are right that what someone wants for a11y reasons
is not 100% predicted by that media query, but the overlap
is high enough that most of the time. And in the rare cases
where the adjustment made by the author actually makes things
worse, users can turn it off.

Only time will tell, but the light-level media-query is
almost certainly the right design for the actual luminosity
use case. The a11y section merely says "UA may offer a manual
switch", which is something we couldn't prevent them from doing
if they wanted to anyway, and just serves as a hint that
while the semantics of this MQ are not perfectly aligned
with a11y use cases, they are close enough that users with
a11y concerns may find this useful.

This isn't even a distortion of the semantics of this MQ.
"light-level:dim" mean that the is too bright for given
the ambient light levels, and the author should tone it down
somehow. "too bright" is subjective, and including not just
the light sensor and the screen type in the equation, but
also taking input from the user seems like reasonable
ting to do for a UA.

This isn't to say that we cannot offer a11y MQs as well,
and I don't think this little section should in anyway
prevent us from also adding a11y focused MQs as well.
But that will not change the fact that adjustments authors
do for washed or dim situations would often be a close
match with the a11y based desires users.

>  c) Deferred light-level to Level 5 to be worked on together
>     with the a11y queries, in order to advance the rest of
>     this draft to CR within the next 6 months, because I
>     really think some of these things are ready and critically
>     needed out in the real world... and the a11y queries are
>     still in the brainstorming stages atm.

It doesn't seem impossible to me that once we're done brainstorming the a11y MQs, we realize that they end up slicing the problem space in a different way, which combines poorly with light-level, and make us wish we'd done it differently.

Not impossible, but not very likely IMO.

 - Florian

Received on Friday, 26 February 2016 10:40:39 UTC