siblingpastry

siblingpastry, 5 days ago (edited 4 days ago)

@aardrian I have the same versions, and I get the same results.

Curious thing though -- if you open and close each expander once, it's fine after that -- the hidden rows are no longer navigable, nor included in the row count.

So for a brutal workaround lol:

document.querySelector('#btnMSa').click();  
document.querySelector('#btnEDENSa').click();  
window.requestAnimationFrame(() => {  
 document.querySelector('#btnMSa').click();  
 document.querySelector('#btnEDENSa').click();  
});  

siblingpastry, 7 days ago

@dgar We call it mean to remind you bloody colonials who’s in charge 🇬🇧⛓️

siblingpastry, 9 days ago (edited 9 days ago)

@VamptVo @SteveFaulkner John's done fantastic work writing-up solution boilerplates for a range of mobile frameworks and situations.

And yeah, the more of them I read, the more I'm just like -- Flutter is bollocks for accessibility. I might start calling it Flubber.

siblingpastry, 9 days ago

@joelanman We treat that as a pass. So long as the difference between required and optional is clear (one way or the other) then it passes 3.3.2.

But required fields still have to be programmatically declared, and my personal opinion is that visual indication patterns should be consistent with that way round.

siblingpastry, 9 days ago

@joelanman If it has visual indication, but no programmatic declaration (required or aria-required) then I’d fail it against 4.1.2.

So yeah, if you’re only marking optional fields with visual labels, and not declaring required fields with those attributes, that’s a pass for 3.3.2 but a fail for 4.1.2

siblingpastry, 9 days ago

@joelanman But there is no state called “optional” and states have to be programmatically communicated. Either it’s “required” or it isn’t, there’s no declarative optional state.

You might be able to make a case that you’re passing the intent if not the letter, but that’s a judgment call, it would depend on the overall situation.

siblingpastry, 9 days ago

@joelanman Ah well, now we’re getting into whether static states are covered by 4.1.2. The general consensus seems to be that they are, and that’s our company stance, but there is debate. I’m AFK right now but there’s a GH thread I’ll find for you later.

siblingpastry, 9 days ago

@joelanman Here's the thread I was referring to -- https://github.com/w3c/wcag/issues/3523

siblingpastry, 9 days ago

@joelanman I think it depends how you read it -- you could find justification in either point of view from the opinions there.

We had this debate recently, and we decided that we would consider static states to be part of 4.1.2.

But often in cases like this, these details form a part of overall decisions -- if an interface is clearly shit, how do we justify failing it without abusing the intent or letter of relevant SCs.

For your case, the greater inclination would be to justify passing it.

siblingpastry, 9 days ago

@joelanman Where are you now -- a mixture?

siblingpastry, 9 days ago

@joelanman We had a recent client who made the same assertion, and that's what made us challenge our standing guidance to mark required fields. From which debate, we told the client, yeah okay that's fine.

But, we still insist on programmatic required attributes.

The inconsistency of that doesn't sit quite right with me. I think that for general usability, you're right, but for AT usability, required indication is better because it's a semantic state. And to my mind, that takes precedence.

siblingpastry, 9 days ago

@joelanman There may also be cases where neither is needed. For something like a login and password form, those fields are always required. That's such an established pattern, you could say that the states are already clear enough that they don't need to be marked.

siblingpastry, 9 days ago

@joelanman It also hinges on how you define "programmatically determinable". I don't really accept that text descriptions count as programmatically determinable, even if they're programmatically associated, but that's a hard hill to die on lol.

siblingpastry, 9 days ago

@yatil @joelanman Only thing with that is the difference between visual and semantic might be confusing for sighted screen reader users. That's the inconsistency that bugs me.

But yeah, I would accept that in conformance terms.

siblingpastry, 9 days ago

@joelanman The impactful difference is for braille output. Braille software uses abbreviations like "req" to reduce verbosity, and it can't do that with text descriptions, because they're not programmatically determinable.

siblingpastry, 9 days ago

@joelanman @yatil It could be yeah, then it's similar to using the hand cursor for buttons -- given that there's a wide variety in real-world use, may as well stick to what's "correct" (ambiguity in that word notwithstanding lol)

siblingpastry, 9 days ago

@yatil @joelanman It just seems slightly odd to me, when visual labels don't match semantic descriptions (or the lack of them). Your example is clear yeah, and maybe I'm the only person who thinks it's odd :-)

siblingpastry, 10 days ago

@joelanman deleteThing or remove if it wants to be a single word.

siblingpastry, 10 days ago

@joelanman nb. you can use reserved words as property names via square bracket notation. Definitely not a good idea though.

siblingpastry, 10 days ago

@matuzo This is too new to rely on. Even Firefox, which implemented it first, has only had support for 12 months. Features younger than 2 years are not suitable for production use.

siblingpastry, 13 days ago

@ollicle Interesting, I hadn't considered using comments. They'd have to be manually cleaned-up though. Could probably use cdata nodes as well.

What do you mean by, play alongside?

siblingpastry, 13 days ago

@ollicle Oh you were asking about dealing with sibling text nodes? Yeah, that :-) Normalizing isn't something you tend to need very often, but it's uniquely useful when you do.

That's a good point about spotting in dev tools, I guess it would depend on what you needed. If the nodes need to be around for a while, it would be more useful to be able to inspect them. My use-case is very transient, they're added and gone again in the space of a synchronous function call.

siblingpastry, 14 days ago

@urlyman It’ll be the same principle, except that conscriptees will be made to do content moderation and social services risk assessments.

siblingpastry, 14 days ago

@urlyman gammontracker.co.uk is available 💡