W3C

– DRAFT –
ARIA Authoring Practices Task Force

25 May 2021

Attendees

Present
carmacleod, CurtBellew, isaacdurazo, jongund, MarkMcCarthy, Matt_King, s3ththompson
Regrets
Bryan Sarah James, Jemma
Chair
Matt King
Scribe
MarkMcCarthy

Meeting minutes

<jesdaigle> +present

[quick roundtable of intros for new folks before scribing begins]

Matt_King: No meeting on June 1st, next meeting is June 8th.

Report on Redesign Project Stakeholder Interview

isaacdurazo: been doing interviews and research over the last few weeks, 12 total - 9 APG users and 3 APG stakeholders

isaacdurazo: learned a lot from stakeholders about the intent/design of APG, learn about the known usability problems and constraints, and what success would look like

isaacdurazo: from users, talked to managers/directors, accessbility engineers, web devs, and some students

isaacdurazo: people with 20+ years of experience and others who were very new. there were also 3 screen reader users in our participant pool

isaacdurazo: i identified that a goal of APG is to be a trusted reference for anyone who has accessibility experience and also needs advice

isaacdurazo: it could point people to other resources, but is mostly for people who already have some knowledge

isaacdurazo: people use APG as a reference, and guide to test widgets against. could be as simple as one component, or a complete web redesign

isaacdurazo: it's also helpful that it's a resource backed by W3C - it has a lot of trust in it.

isaacdurazo: i found 3 levels of usage for certain sections - _usage_, not necessarily importance.

isaacdurazo: description of design patters and widgets is what people use the most. this also includes example widgets and pages, mostly because they're easy to consume and digest

isaacdurazo: 2nd most common is the introduction (readme, warnings, etc.), and things like info about landmarks, accname, and description

isaacdurazo: finally, developing keyboard interfaces and how to use semantics and related regions. i put this last because people use it, but couldn't always remember a case where they needed it

isaacdurazo: some other highlights - "No ARIA is better than Bad ARIA", many people like having this up front. some people even wished it was more prominent

isaacdurazo: another one was the keyboard interaction table, roles/prop/state tables - lots of positive comments about this content

isaacdurazo: i also documented frustrations. pretty unanimously, people think APG is a bit too long, overwhelming, sometimes too verbose

isaacdurazo: people mentioned getting lost, not knowing when you transition from one pattern to another.

isaacdurazo: people also thought there is some information missing, like mobile patterns, touch support (which APG does state isn't included, but people would like it)

isaacdurazo: a few participants considered some of the examples and patterns not being very representative of real world cases (maybe 2 or 3 participants)

isaacdurazo: these people had some ideas, but don't find it easy to engage with the community group

isaacdurazo: so that might be another area to improve, that's not about the APG document itself

isaacdurazo: people shared that having a central nav would be nice, rather than a list on the side. most agreed content should be broken up more.

isaacdurazo: many shared that a sorting/filtering/search mechanism would be helpful, as well as jumping right into the example content.

isaacdurazo: some people wanted a "simple english" type of writing style, some wanted an indicator of what's recently changed.

isaacdurazo: having a better experience for mobile devices came up, as well as expandable/collapsible areas. the ability to embed examples or pieces of APG elsewhere came up as well

isaacdurazo: some older folks (having used APG since the beginning in some cases!) expressed a desire to still have the one-page view

isaacdurazo: having an easier way to participate, submit feedback or ideas, or generally interact w/ the group was mentioned often.

isaacdurazo: with regard to priorities, the biggest one is going to be improing the UX. first step is probably going to be coming up with a new information architecture

isaacdurazo: that should help us implement filtering/searching, then that'd lead us to filling in remaining gaps (like tutorials, mobile patterns, etc.)

isaacdurazo: improving the UX and adding these features, while also using the Ed and Outreach group's template, should make the APG a lot stronger and build up its trust and reliability

isaacdurazo: happy to answer any questions or anything else for anyone

Matt_King: thank you so much Isaac!

Matt_King: we'll get the report published and up once we get the repo set up

carmacleod: you mentioned some examples didn't have good real world cases, do you know which ones those are?

isaacdurazo: this report is pretty general, but i can look at the notes if they mentioned it. i didn't want to dive too deep into how the content was written though.

Matt_King: i had a similar question, but I want to be careful that we're getting that feedback from a broad base

Matt_King: which makes me excited to think of a way for readers to give even super lightweight feedback (like is this page helpful/not helpful) _in addition_ to filing an issue

isaacdurazo: it's important to acknowledge that - even when we get feedback like that, it might not be true for everyone

Matt_King: I'm just as curious as carmacleod though, so if you can find that!

isaacdurazo: i'll see what i can dig up!

Matt_King: other questions/comments?

jongund: did you find any differences between what screen reader users said that was specifically different from others?

isaacdurazo: yes - one said he uses APG for educational reasons, and he finds it too cumbersome to use (though it's helpful for his students).

isaacdurazo: another said he's used to it, but didn't necessarily complain about it. might be that he's experienced with APG's layout.

isaacdurazo: even others said things like "it's not the best, but i'm used to it"

Matt_King: as a screen reader user, i don't think about getting lost that much, and it's difficult for me to scroll into something separate. but I can understand how a visual user would get lost

Matt_King: so what's next isaacdurazo?

isaacdurazo: next, i hope to start putting together proposals for a new architecture, maybe over the next 3-4 weeks. i'll get some wireframes made, then get feedback

isaacdurazo: i'm going to start on those wireframes this week, and maybe i'll get something shareable by the end of the week.

Matt_King: would you be ready to share on Jun 8 or 15?

isaacdurazo: that's the plan

isaacdurazo: thanks everyone!

Update on Merging Slider Examples

Matt_King: we've merged temp and rating sliders, next up is seek

Matt_King: looks like the seek slider needs functional, visual design reviews; a11y on MacOS and Android

carmacleod: I didn't see that I was assigned when I looked last, I'll get these sorted after the call

Matt_King: we'll chat with Sarah, Jemma said she'd wrap up her piece later this week. Siri, is your testing done?

Matt_King: oh wait - I see your approval, nevermind!

Matt_King: so with any luck we'll get this one in within a couple days.

Matt_King: multithumb slider - Siri, do you approve? I see your testing but not an approval

Matt_King: jesdaigle, you reviewed this a while ago, but jongund made changes since then. also some linting errors, jongund. jesdaigle, can you re-review after Jon resolves those errors?

jesdaigle: yep

Matt_King: the linting errors appeared recently, maybe there was an update to the linter or something.

Matt_King: so our Range thermostat, based more on HTML... seems to me it'd be effective to say "here's an ARIA version, but you can do the same with MUCH LESS ARIA, and simple HTML."

Matt_King: thoughts from everyone? If we agree that HTML is a good thing to extoll, which other one should we do with HTML?

Matt_King: would be nice to do one that doesn't use JavaScript

jongund: James felt it would've been useful to show ranges etc. JS can also be helpful for testing to make sure everything is working right

Matt_King: would it be possible to make the color viewer with HTML?

jongund: it'd probably be pretty simple, yeah

jongund: but we change the color of the rail as things move around, which was part of it. so we might still need some scripting'

Matt_King: i guess we could leave out keyboard handling, and if there were differences, that'd be worth noting, right?

jongund: if we're just doing one slider, there might be only some minor differences. scripting seems kind of minimal.

Matt_King: i think we could also use the same ARIA-AT tests for both, which would be cool

Matt_King: any other thoughts on the idea of doing HTML versions? The idea being to demonstrate how powerful HTML and CSS are, even without ARIA

jongund: idea being ARIA can supplement, vs. be all or nothing

Matt_King: right

Review Jumpto Content Implementation

Matt_King: i'm not 100% sure how we want to approach reviews here; jongund has addressed many comments and concerns already

Matt_King: want to make sure we're having it show up on all the pages we need it to

Matt_King: having things show up twice in the menu doesn't serve much purpose, so maybe we should address that?

Matt_King: but that's a small thing, only shows up on the index page

<siri> https://github.com/w3c/aria-practices/pull/1860#issuecomment-840617404

carmacleod: on every example page, the nav at the bottom itself might not be useful anyway

carmacleod: our validation tool catches it all the time too because it doesn't have a name

jongund: maybe if we don't remove that bottom nav element, we could put it in contentinfo?

Matt_King: or maybe change the role? either way, that's a separate issue maybe

Matt_King: it'd also be a manual change to every example page, unfortunately

jongund: things like this can be useful because these things become apparent when adding features like jumpTo

Matt_King: so it seems like the access key (alt+0) gets announced twice by JAWS, do you know why?

jongund: might be the tooltip with aria-describedby

Matt_King: so if we take that out, what's the other source of announcement?

jongund: the access key

Matt_King: if there's some screen readers that don't announce an access key, then we want to keep that. so we'll need to look at NVDA (I'll do that)

jongund: i fixed the multithumb linting error from prior issue, while we've been chatting

siri: so what's the decision on these navigation landmarks?

Matt_King: this only shows up on one page, so maybe we change it on the index page then

Matt_King: this might have to be a different PR Siri

jongund: well it's configurable, so we could remove the last one

Matt_King: we could consider having it not show the navigation the bottom of the page, that might be easier too

jongund: okay i'll work on that

Minutes manually created (not a transcript), formatted by scribe.perl version 131 (Sat Apr 24 15:23:43 2021 UTC).

Diagnostics

Succeeded: s/No meeting on June 1st, next meeting is June 8th./Matt_King: No meeting on June 1st, next meeting is June 8th.

Succeeded: s/2nd is the/this also includes

Succeeded: s/color slider/color viewer

Maybe present: jesdaigle, siri