Re: Target spacing refinement

Hey Alastair,
I think this variation of Detlev's solution works:

*For all adjacent targets, the distance from the farthest point of one
target is at least 24 CSS pixels away from (every point on) the other
target, except if:*


   - *Diameter: The smallest diameter is at least 24 CSS pixels;*
   - *... *


I've gone through 50-ish test cases (
https://codepen.io/wilcofiers/pen/abZxPow), and this works for all of them.
This changes the existing suggestion in the following says:

- By using "all adjacent targets", it becomes clearer that it's comparing
everything to everything else nearby. Hope this is sufficiently clear. Had
this been ACT we'd probably have phrased it "For every ordered pair of
distinct targets", but that's probably a little too "mathematical".
- Changed "any point" to "every point", any sounds like you get to pick. I
put the phrase in brackets because I'm not convinced it's needed.
- By starting at "farthest point" and ending on "every point" we flip the
direction around, so that we're failing the right target.
- Replace "farthest edge" with "farthest point", since you can't measure
the distance from a point to a line in a single number.
- Replace "Nested" by "Diameter": This exception is also needed for partial
overlap (see example G3 and G4). Anything that's 24px in diameter gets a
pass. By using diameter instead of width and height, it accounts for odd
shapes like stars where the smallest diameter is far smaller then the
width/height.


On Wed, Nov 18, 2020 at 5:15 PM Wilco Fiers <wilco.fiers@deque.com> wrote:

> Hey folks,
> I did what I always do when rules get too complex. I write test cases.
> Here's what I wrote. I used color gradients to indicate passes and fails.
> Light green to dark green is passed, dark red to pink is fail.
> https://codepen.io/wilcofiers/pen/abZxPow
>
> Unfortunately, none of the proposals actually gets all of them right. So
> this is going to need more work. I'll see if I can come up with a proposal
> that gets all cases right. Probably worth for folks to have a look, see if
> we're all in agreement on these. Maybe most noteworthy are E3 and E4. Those
> corner blocks pass with the currently published SC text, but they fail in
> all of the new .
>
>
>
> On Wed, Nov 18, 2020 at 4:41 PM Alastair Campbell <acampbell@nomensa.com>
> wrote:
>
>> Hi Michael,
>>
>>
>>
>> Tackling the second one:
>>
>> > *The distance from each target's mid-point to the mid-point of
>> adjacent targets is at least 24 CSS pixels, expect when...*
>>
>>
>>
>> Measuring from mid-points allows for tiny targets next to larger ones,
>> e.g:
>>
>>
>>
>> Although easier to understand (slightly), I don’t think it aligns to the
>> goal quite as well.
>>
>>
>>
>> For the re-write of option 5, I think it would need to start with the
>> thing you are evaluating, e.g:
>>
>> *For each target, the distance from each adjacent target to the farthest
>> edge of the current target is at least 24 CSS pixels except when:*
>>
>>
>>
>> If others think that scans ok, I’m happy with that.
>>
>>
>>
>> Regarding the ‘objectives’, I think we can easily include that on the new
>> understanding docs at the top of the intent, and work back through the
>> 2.1/2.0 docs later.
>>
>> The upcoming re-design looks like this for the understanding doc:
>>
>>
>> https://w3c.github.io/wai-wcag-supporting-documents-redesign/2020-07-15-prototype-understanding.html
>>
>>
>>
>> We can add a CSS class to the objective paragraph and work out the
>> styling in parallel.
>>
>>
>>
>> Cheers,
>>
>>
>>
>> -Alastair
>>
>>
>>
>>
>>
>> *From:* Michael Gower
>>
>>
>>
>> I agree option 5 *seems *to scan best, but I also think there is a
>> missing preposition. There are 2 ideas here:
>> 1) we are talking about the edge *farthest from* an adjacent target
>> 2) we are talking about the distance* from *that edge *to* the adjacent
>> target (or *between *them)
>>
>> So I think we need 2 prepositions, one to describe which edge and one to
>> describe the distance *between* two points. i think a rejig of the
>> sentence still allows that to scan okay:
>> *The distance from each adjacent target to the farthest edge of the
>> current target is at least 24 CSS pixels.*
>>
>> I think we need to bear in mind that this is a design-centric
>> consideration. As such, it is even more important to get the
>> language/concept simple. As such, I want to advocate for a variation I
>> pasted into the channel yesterday:
>>
>> *The distance from each target's mid-point to the mid-point of adjacent
>> targets is at least 24 CSS pixels, expect when...*
>>
>> AWK said that this wouldn't work for some edge cases, but I'd like to see
>> some examples to understand what gets through the net.
>>
>> Regardless of wording, this is another SC where a quick blurb summarizing
>> the objective would help with rapid comprehension. For instance:
>> *Objective: Ensure separation of targets for ease of operation.*
>> I wrote such blurbs for all the 2.1 additions, which were supposed to be
>> included in the Understanding documents, but were never incorporated.
>>
>> Michael Gower
>> Senior Consultant in Accessibility
>> IBM Design
>>
>>
>> 1803 Douglas Street, Victoria, BC  V8T 5C3
>> gowerm@ca.ibm.com
>> cellular: (250) 661-0098 *  fax: (250) 220-8034
>>
>>
>>
>> From:        Alastair Campbell <acampbell@nomensa.com>
>> To:        "WCAG list (w3c-wai-gl@w3.org)" <w3c-wai-gl@w3.org>
>> Date:        2020/11/17 04:34 PM
>> Subject:        [EXTERNAL] Target spacing refinement
>> ------------------------------
>>
>>
>>
>>
>> Hi everyone, After the long discussion on target spacing today,...
>>
>>
>>
>>
>> *This Message Is From an External Sender*
>>
>> This message came from outside your organization.
>>
>> Hi everyone,
>>
>>
>>
>> After the long discussion on target spacing today, I tried to collate the
>> options into one place and add a couple of diagrams:
>>
>>
>> https://docs.google.com/document/d/1Q9zWT1OjdCrts2xuadVEaJ2wpyLzxnysFQCSTs72L2o/edit?usp=sharing
>>
>>
>>
>> Personally, I’m leaning towards option 5 as the simplest which measures
>> the size+spacing of the target, which would be:
>>
>>
>>
>> For each target, the distance of the target’s edge farthest from each
>> adjacent target is at least 24 CSS pixels, except when:
>>
>>    - [3 bullets unchanged]
>>    - *Nested:* The target is enclosed within another target and has a
>>    minimum height and width of 24 CSS pixels.
>>
>>
>>
>> If you’d like to add something (options, positives/negatives, diagrams
>> etc) please let me know and I’ll add you as an editor of the doc. It is
>> open for comments.
>>
>>
>>
>> Kind regards,
>>
>>
>>
>> -Alastair
>>
>>
>>
>> --
>>
>>
>>
>> @alastc / www.nomensa.com
>>
>>
>>
>
>
> --
> *Wilco Fiers*
> Axe-core product owner - Co-facilitator WCAG-ACT - Chair ACT-R
>
>
> Join me at axe-con <http://deque.com/axe-con> 2021: a free digital
> accessibility conference.
>


-- 
*Wilco Fiers*
Axe-core product owner - Co-facilitator WCAG-ACT - Chair ACT-R


Join me at axe-con <http://deque.com/axe-con> 2021: a free digital
accessibility conference.