@Lukew@toot.wales

· 3w ago

It looks like Google is finally working on a way to support third-party browsers being able to install proper APKs for web apps. This is generally something that only Chrome can do currently. There's very little public documentation and it still seems quite early, but there's a new WebAppManager API added in Android 17. https://developer.android.com/reference/android/content/pm/webapp/WebAppManager

developer.android.com

15

1

7

0

Luke @Lukew@toot.wales

· 3w ago

Replying to

@brucelawson@social.vivaldi.net alternatively you were very successful! Going through the EUs processes to ensure this gets across the line in a satisfactory way is still good to do. As I say at the end of the thread there's still rough edges that will need cleaning up.

2

0

Luke @Lukew@toot.wales

· 7mo ago

People at OpenUI have been working on a revamped implementation of focusgroup.

Please go try it out, check it does what you expect! Provide feedback, let us know if it's broken.

I'm especially interested to know how the accessibility is for it. Thought has been put in to ensure behaviours have semantics attached.

https://developer.chrome.com/blog/focusgroup-rfc

developer.chrome.com

10

0

11

0

Luke @Lukew@toot.wales

· 5mo ago

As of latest Safari Tech Preview (242) support for the dialog closedby attribute is enabled (not enabled for stable yet). If people could give it a test that would be a great help!

https://developer.mozilla.org/en-US/docs/Web/API/HTMLDialogElement/closedBy

developer.mozilla.org

5

2

3

0

Luke @Lukew@toot.wales

· 5mo ago

Replying to

@jyasskin@hachyderm.io @jaffathecake@mastodon.social @slightlyoff@toot.cafe alternatively that's developers with legitimate opinions only just becoming aware of this? I get individual and aggregate emails of intents on blink-dev and hasn't spotted the i2s was even filed, now think about developers who have a day jobs where that's just not relevant, or who don't even know where to look for standards stuff. Obviously it could be a "mob" but I think it's unfair to completely disregard as one

3

0

Luke @Lukew@toot.wales

· 5mo ago

Replying to

@nachtfunke @jaffathecake I think the big power developers do still hold is to just not use this API. The web isn't cast in stone (as much as people may think it is), if the usage stays low enough there's a chance it might unship in years to come, or at least go through changes to make others more onboard (idk if that's feasible here though).

3

0

2

0

Luke @Lukew@toot.wales

· 7mo ago

Replying to

@keithamus@indieweb.social @joface@mastodon.scot I pulled out a cloth and wiped my screen down to get the optimal score I could 😂 glad it's not just me

4

2

0

Luke @Lukew@toot.wales

· 5mo ago

Replying to

@_r @servo have you raised an issue about this? Is there a specific page? Loading it in servo on macOS it seems to load okay?

2

1

0

Luke @Lukew@toot.wales

· 6mo ago

Replying to

@joshtumath @keithamus okay I tried again and got 62/80 this time.

2

0

Luke @Lukew@toot.wales

· 6mo ago

Replying to

@ar1 @WPalant @freddy did you miss the "otherwise more apps and platforms will follow." You don't use meta apps, great good for you, but it's still a dangerous move.

2

9

0

Luke @Lukew@toot.wales

· 7mo ago

Replying to

@dev @freddy Yeah I suspect DOMPurify is the best fallback approach rather than an attempt at a true polyfill.

2

1

0

Luke @Lukew@toot.wales

· 5mo ago

Replying to

@knowler@sunny.garden that iOS behaviour is explicitly *not* part of this. The implementation currently is focused on getting light dismiss included and enabling or disabiling the existing escape key press. The back gesture (if it's ever included) will be part of the close watcher API behaviour which this can integrate into.

1

0

Luke @Lukew@toot.wales

· 5mo ago

Replying to

@knowler@sunny.garden glad to hear! I think the main thing that's not quite right (but is an existing issue) is the behaviour of stacks of dialogs and popovers and perhaps even full screen elements. I'm hoping to see if we can get closedby shipped even if that side of things is less developed. I expect 99% of usage is just slapping closedby=any on a modal where you won't notice any difference in that area.

1

0

Luke @Lukew@toot.wales

· 5mo ago

Replying to

@freddy I'm curious if it's allowed the team to rethink certain architectural decisions to better guard against certain vulnerabilities across the board? Or if it is still a case by case bases where bugs get found but can also still sneak in?

1

0

Luke @Lukew@toot.wales

· 8mo ago

Replying to

@joshtumath@mastodon.social @jscholes@dragonscave.space @mahryekuh@hachyderm.io for something like color scheme if we had made that honour dark mode all of a sudden it would have caused a lot of breakage and led to very real accessibility issues (contrast bugs galore). Whereas in this case it does seem to be opt-in already (by virtue of having changed that font-size setting)?

2

0

Luke @Lukew@toot.wales

· 5mo ago

Replying to

@_r @servo okay well I've just tested and that seems to work on macOS on 0.1.0. So it's either fixed or it is potentially platform specific. Completely understand if you don't have time or energy for it, but fwiw a submission as simple as that URL, your OS and version and the crash log you (presumably) get from the tab is a great help.

1

0

Luke @Lukew@toot.wales

· 5mo ago

Replying to

@manul70 @servo I've just raised a bug for this, it seems they rely on the web animations API in a critical path, and servo doesn't support this.

1

0

Luke @Lukew@toot.wales

· 6mo ago

Replying to

@joshtumath @keithamus some of the long hands get me.

1

0

Luke @Lukew@toot.wales

· 6mo ago

Replying to

@jaffathecake cursed, I like it.

1

0

Luke @Lukew@toot.wales

· 6mo ago

Replying to

@mez@mastodon.nz @leaverou@front-end.social I had the exact same thought process, "what why would it be different in the last case??, hmm okay maybe I'm missing something, but B seems right... Or maybe A 🤔" voted B too.

1

0

Luke @Lukew@toot.wales

· 6mo ago

Replying to

@matuzo @sarajw @aloxe ahh I see, yeah it doesn't help lots of people say it the wrong way too so it's hard to actually know.

1

4

0

Luke @Lukew@toot.wales

· 6mo ago

Replying to

@sarajw @matuzo @aloxe it's definitely said like veet, because it's french for quickly!

1

7

0

Luke @Lukew@toot.wales

· 8mo ago

Replying to

@joshtumath@mastodon.social @jscholes@dragonscave.space to end positively though, thank you for your work to get us to even this stage! A solution really is better than no solution!

1

2

0

Luke @Lukew@toot.wales

· 8mo ago

Replying to

@joshtumath@mastodon.social @jscholes@dragonscave.space This is where my disagreement is (please don't take this to be argumentative I don't intend for that at all) but the situation today for people with text scaling *is* broken and unreliable. They've asked for big text and they aren't getting it. If this is opt-in I think we really should rally to get sites to adopt it. Lighthouse should flag it as missing. Eventually Google should flag it as an negative for indexing (they already do if you have too small click targets).

1

2

0

Luke @Lukew@toot.wales

· 8mo ago

Replying to

@joshtumath@mastodon.social @jscholes@dragonscave.space @mahryekuh@hachyderm.io

Yes, because those sites are already broken, we can't change how browsers on iOS and Android scale text by default

Why, can't we change that? The answer currently seems to be because that might break assumptions and make rendering "bad".

I believe James' point was: why is potentially broken layout inherently worse than too small text?

The 'breakage' as I understand it would only happen for users who have changed their font size (a conscious a11y choice)

1

2

0

Luke @Lukew@toot.wales

· 6mo ago

Replying to

@bkardell oh yeah that's definitely a thing. I find it when I'm very hungry if I eat a little bit but not enough to fill me I just end up feeling worse.

0

4

0

Luke @Lukew@toot.wales

· 6mo ago

Replying to

@jaffathecake that seems very broken in chrome canary on android at least. The select picker never opens but the values randomly change.

0

4

0

Luke @Lukew@toot.wales

· 8mo ago

Replying to

@brucelawson@social.vivaldi.net doesn't have extensions of its own? But it can run chromium extensions right? If so why does it need its own extensions?

0

3

0

Luke @Lukew@toot.wales

· 8mo ago

Replying to

@joshtumath@mastodon.social @jscholes@dragonscave.space IDEs new HTML file scaffold should include it, etc etc. Obviously this isn't something you can control yourself just saying this is what we need to make this work for users. I fear the situation will be that some small subset will use it. The rest won't, and it'll be users trying to report site bugs and getting ignored rather than potentially sites reporting browser bugs and hopefully getting told hey fix your shit.

0

2

0

Luke @Lukew@toot.wales

· 7mo ago

Replying to

@christianp i was saying this to Keith! I'm colour blind but my eyes could see better than his cut off point (for certain colour hues). So he gamified it. I have 0.0044 on my Mac and 0.0066 on my pixel phone.

0

Luke @Lukew@toot.wales

· 5mo ago

Replying to

@matuzo@front-end.social if I see one more circular checkbox I swear...

0

Luke @Lukew@toot.wales

· 6mo ago

Replying to

@simounet @jaffathecake https://open-ui.org/components/combobox.explainer/#making-your-own-picker-for-a-filterable-select - I believe this is what you're looking for, assuming you want to filter a select. Alternative the input+datalist approach mentioned in the same explainer does the alternative type of combobox.

open-ui.org

0

Luke @Lukew@toot.wales

· 8mo ago

Replying to

@joshtumath@mastodon.social need to tweak my paddings a bit but even without that it works pretty well on my site!

0

2

0

Luke @Lukew@toot.wales

· 6mo ago

Replying to

@jaffathecake yes, experimental web platform features. Disabiling that fixes it.

0

2

0

Luke @Lukew@toot.wales

· 7mo ago

Replying to

@freddy @dev is there any attempt at a polyfill? I guess it's not really doable. Something using trusted types could work I suppose but could still have mXSS issues.

0

3

0

Luke @Lukew@toot.wales

· 8mo ago

Replying to

@joshtumath@mastodon.social @jscholes@dragonscave.space coming back to this I am a little worried that this is something we'll come to regret as yet another magical meta tag you have to add for your page to just do the right thing. Except it's less obvious when it's missing than say the viewport meta tag. I know "don't break the web" is the theme of standards but I really do wonder if in this scenario it's the right thing to do. Maybe a meta tag to opt out if you absolutely must?

0

2

0