Re: User Agents Do Not Implement Absolute Length Units, Places Responsive Design in Jeopardy

*10pt text is unreadable on a monitor at TV-distance viewing, and is
excessively large on a hand-held device with a high-res screen capable of
displaying crisp text at smaller sizes.*

David, this is exactly why factual physical units are really, really
important and useful.  A screen that will be viewed at TV-distance is likely
a large display (e.g. a TV).  In this situation, media queries would make
said text larger as we can safely make this assumption (larger screen,
larger distance) in nearly all cases.

We can make all kinds of accurate assumptions in this way.  A 3"-wide screen
is probably a touchscreen phone and should have big, finger-friendly UI
elements (we can enhance this prediction by detecting touch events).

You make a case by mentioning a far-future technology like wall screens that
are feasibly 10 or more feet wide, but the same case could be made for
non-rectangular or holographic displays to undermine anything in CSS.  The
supposition that subtended units are more useful for the lion's share of
actual web design simply isn't true.  I wouldn't debate their utility, but I
think the goal for units like "cm" should be to abolish old ambiguities, not
introduce a new one.

-Brian


On Fri, Oct 21, 2011 at 5:47 PM, David Singer <singer@apple.com> wrote:

>
> On Oct 21, 2011, at 13:10 , Brian Blakely wrote:
>
> This is an idea I strongly believe should be a separate unit or solution
> path.
>
> It's super-useful for something like Sony's wearable display<http://technabob.com/blog/2011/09/04/sonys-hmz-t1-3d-oled-goggles/>,
> wherein the viewport is physically small but apparently (I haven't tried it)
> functionally equivalent to a big-screen display.
>
> However, I feel that is a corner case, and in most cases authors will want
> to use linear physical data, to then arrive at their own design conclusions
> about legibility and appropriate use.
>
>
> I'd really encourage you to scan the archives, if you haven't already,
> because we had a long discussion of this.  Basically, in days of yore, when
> computers were real computers etc., the web was browsed on computer screens
> at about the distance and size of a piece of paper.  Using physical
> measurements like 10pt etc. 'worked'.  10pt text is unreadable on a monitor
> at TV-distance viewing, and is excessively large on a hand-held device with
> a high-res screen capable of displaying crisp text at smaller sizes. It's
> meaningless on projection glasses, and unimplementable on jumbotrons (whose
> pixels tend to be of the size of inches).
>
> The use cases for true physical seem to be when you're comparing or using
> the page as a physical embodiment; you're using your tablet as a carpentry
> template, for example, and that one-inch hole you're making in the door
> should really be one inch or the lock won't fit.
>
> Whole-wall displays are also getting more common, which have a size not
> unlike a jumbotron, but a vastly different viewing distance, subtended
> angle, and usage.  Mere physical size would again be misleading, I think.
>
> The old comforts are fast disappearing, I fear.
>
>
> Cheers,
> -Brian
>
>
> On Fri, Oct 21, 2011 at 2:28 PM, David Singer <singer@apple.com> wrote:
>
>> Perhaps size detection should be based on the angle subtended at the eye?
>>  That's what we are basing other stuff on.  I think the linear size is much
>> less relevant (and meaningless in some cases).
>>
>> To do it well, you probably need to know "how much is subtended by the
>> display" .  You probably should not need to know "what is the pixel count in
>> that range" as the equipment designer should have made it that 'normal'
>> sizes like 10pt text are, in fact, readable.
>>
>> On Oct 5, 2011, at 10:28 , Boris Zbarsky wrote:
>>
>> > On 10/5/11 1:21 PM, Brian Blakely wrote:
>> >> The problem isn't resolution detection.  The problem is that we're
>> using
>> >> pixel resolution to do *size *detection.  Adjusting a layout for
>> >> small-size displays is currently achieved by defining max-width MQs at
>> >> about 640 *pixels*, when it really should be 3.5 *inches* (ballpark).
>> >
>> > The problem is that to be web-compatible, UAs more or less have to fix
>> 1in == 96px.
>> >
>> > And again, just to be clear, are you trying to detect _linear_ size, or
>> _angular_ size of the screen?
>> >
>> > -Boris
>> >
>>
>> David Singer
>> Multimedia and Software Standards, Apple Inc.
>>
>>
>>
>
>  David Singer
> Multimedia and Software Standards, Apple Inc.
>
>