Few seats left for our Advanced Backstage Plugin Development workshop on December 7th. Secure your seat today!

111: Accessibility in Single Page Applications

Hosted byCharles Lowell, Wil Wilsman, Robert DeLuca

September 20th, 2018.

In this episode, Robert, Charles, and Wil talk about the whys and hows of accessibility, as well as what makes single page applications special, why they are they harder for accessibility, and frameworks that can do this for you.


This show was produced by Mandy Moore, aka @therubyrep of DevReps, LLC.


ROBERT: Hello everyone. Welcome to The Frontside Podcast. This is Episode 111. I'm Robert DeLuca, a software developer here at the Frontside and I'll be your episode host. Today, we're going to be discussing accessibility in single page apps. With me as co-hosts are Charles Lowell. Hey, Charles.

CHARLES: What is up Robert?

ROBERT: And Wil Wilsman. Hey, Wil.

WIL: Yo, yo, yo, yo.

ROBERT: Sounds like we're ready to drop a disc track. We're not going to be dissing anybody here. We're going to be talking about helpful things with accessibility in single page apps. Before we get into the nitty-gritty of accessibility in single page apps because we're getting into some deep stuff, I think I want to cover a lot of 'how' because I know accessibility things are usually about why you should be doing it and then they touch on things like, "You should be using alt attributes for your images," but for single page apps, I think we need to go further.

CHARLES: It always ends up like, "Then draw the rest of the accessibility owl."

ROBERT: Yeah, here's your two circles and the alt-tags are your circles and then the rest of the freaking owl is focus management and everything else that comes with it. Before we get there, what is accessibility? I guess, if we trim the giant umbrella down a little bit from everything that is accessibility that can be physical space things, like wheelchair accessible ramps or things like that, what about just technology?

WIL: People need assistive tech to interact with technology such as switches or keyboards and obviously screen reader is a big one but when it comes to the software itself, you could even talk about colors and people who are color blind, so red might not be red to everybody.

ROBERT: Speaking from experience there?

WIL: Yeah. I'm colorblind. Red is brown to me.

CHARLES: Things like hearing and all of it, right? It really is like just designing it in such a way that it can be used by as many people as possible.

ROBERT: Right.

WIL: And that includes your mom who may not be the best with technology but she still needs to pay your bills online or something.

ROBERT: Exactly, yeah and in for context to listeners that might not know, my mom is 100% blind, so it's kind of where it comes from.

CHARLES: But my mom is not but she has all kinds of problems.

WIL: Yeah, same with my mom.

CHARLES: That also falls under the category of accessibility, right?

ROBERT: Absolutely.

CHARLES: Right, accounting for age and culture.

ROBERT: We're blending into the why of accessibility, which is perfect. One of the things that it's such a good segue because what people are starting to realize and I think why accessibility is really starting to catch wind and get some traction is because a lot of people that grew up in the technology age were open in using technology a lot. Our parents probably did not use technology heavily. That's definitely the case for my dad. He still has a flip phone and he says, he's a low tech man living in a high tech world and just refuses to pick up technology but those are the people that just didn't use technology. But now we have a lot of people that grew up with technology and use a lot and they're aging into more disabilities and they're going to need that accessibility, which I think is really interesting to think about because that's a lot of buying power if you're just going to start moving in that needs in accessibility, right?

CHARLES: That's true. I know I may need glasses pretty soon, so colors and fonts were going to be heck a lot more to me in the next five years than they have over the last 20.

ROBERT: Yup, exactly and that's going to be huge for that and that's one piece of the why, so what are the other reasons that you'd pick up accessibility other than people saying that it's morally correct. I don't like starting off conversations for accessibility because it's the thing that you should be doing.

WIL: I think it goes even to my user experience like power users that don't like using the mice or mouse. That's me. I really prefer to just use my keyboard for everything. When the new Firefox browser came out and I couldn't navigate through the menus the way I was used to, I went back to Chrome.

ROBERT: That's interesting.

CHARLES: I still have not found a good workflow for navigating tabs with the keyboard without just kind of twisting my wrist all out of shape. You have to share that with me but again, that's another thing. That's an impediment that sits between me and the application, that actually --

ROBERT: Which is really interesting, you're getting into a keyboard navigation and focus management, which is really the crux of accessibility for screen reader users and switch users.

CHARLES: What I'm hearing is that in this case, including good keyboard and focus management in your application, e.g. making it accessible to screen readers, you at the same time, enabling your power users. I think a great point is that by introducing these very low friction workflows, you're actually going to be enabling other parts of your customer base and not just catering to one, that there are ripple effects throughout your system.

ROBERT: It may set it for everyone.

WIL: Yeah, not only people who need it but people who don't know they need it.

ROBERT: Yes, exactly. Think about the WCAG spec as user experience guidelines. They're not telling you how to implement a thing specifically for a screen reader. They're telling you how to implement it in a way that works for everyone, regardless of what ailment they have. It could be a temporary ailment. It could be a permanent ailment, where they have to use a screen reader or any kind of ailment that you can think of. They're not considering just one use case. It's a broad thing that shows how you can make your application better for everyone. I think that's a better way to look at the WCAG spec than I need to read through this and make sure that this auto complete works for a screen reader.

If you look at the guidelines, they're not telling you just first screen reader. They're telling you how to make it work for a switch, someone who is colorblind or who is using a dictation software. I kind of tend to look at that as UX guideline, which really helps me build a better app overall because when you nail down that user flow, everyone benefits from it because it's pretty well thought out at that point.

CHARLES: I like that too and I think that thinking of it as user experience and making sure that you have a complete user experience is a good way to think about it because it kind of separates the concerns of, "I've got HTML but is my application really HTML or is it a set of workflows and the data over which those workflows operate?" It really forces you to think my application is not a set of React opponents or web components or Ember components or what have you but really, there's a deep structure to it and it makes you kind of shine a light on that deep structure and try to map its surface. Then, if you really, really know it and you capture it, then you can represent it in any medium.

I think it's just a win not just for one niche group of users but also for all of our users and then also for your future users that you don't have because your application is designed better and is going to work. Who knows? Maybe there's some new interface or some new medium, some new device that comes out that hasn't even been invented yet. But if you really have a strong internal representation of what your application is, you're going to be able to be the first to move to that market.

ROBERT: Right and like I said earlier, people are starting to aid in the needing accessibility thing. If you need a dollar to justify it, that's going to be a big reason coming up. As a part of the new WCAG 2.1 spec, there is a zoom. I forget what this access criteria number is but the new criteria says your app basically needs to be responsive. That kind of maps directly to what you said earlier, Charles which is like you're going to need glasses soon and fonts and being able to zoom. It's going to be really important.

We see that just pop up everywhere. To give a counter example of not just for accessibility like you need it for glasses but the other side of that could be like when we give demos on low resolution projectors or screencast or anything, when we zoom the screen, the apps just still be usable and you should still be able to demo it and that's just something that you have both sides to that, where accessibility kind of works out for everybody. People on the call probably don't need glasses but to see that tiny screen that's being shared, you should be able to assume that.

CHARLES: Right. I'm just kind of restating what you said but I want to make sure to call it out explicitly because it was a definitely an aha moment for me, that you basically if you build yourself an accessible website, you build it so that it works on projectors and mobile devices too, so you kind of killed two birds with one stone and you don't have to make a special effort because zooming in the screen is tantamount to viewing it on a phone or viewing it on a tablet.

WIL: Yeah, we mentioned it earlier about accessibility and physical space and one of those things is being able to access something from anywhere no matter the device that person is using.

ROBERT: Yes. That's a lot of the 'why' of accessibility and I think we did a pretty good job of staying away from the more argument because morally we don't know if it's the right thing to do but a lot of the times it's not in front of you, it's hard to do and I want to make sure that you don't feel bad if you're not building accessibility in your apps. It's not easy. I don't sort of like people get up and say that, "Accessibility is easy. Just do this," because it's not --

WIL: If it was easy, everybody would do it.

ROBERT: Exactly. I always come back to accessibility being just like UX -- user experience -- because if it were easy, everybody would have a really great user experience too. It's a hard thing to boil down and to simplify it, right?

CHARLES: Right and like every other aspect, the kind of nonfunctional requirement, I say nonfunctional requirement that's a little bit of a contradiction terms but it becomes very hard if you haven't done it from the beginning. But if you didn't start with an accessible app, you weren't thinking about that or you inherited the app or it just wasn't on your radar for whatever reason. If you got a codebase that's two years old, going back and trying to make it accessible, it was extremely hard. It's expensive, expensive, expense.

WIL: Yeah, it's going to cost more in terms of money and time to added it after the fact.

ROBERT: Exactly. I spent a year on helping on Visa Checkout and there were some accessibility considered in the designs but a lot of the time my feedback wasn't just like, "Yeah, sprinkle some ARIA attributes on there and you're good." It was like, "Do we really need a carousel here to represent your list of carts because it's really hard to navigate around that?" A lot of the times, it ends up boiling up to is this the best way we can represent this data? Is this the best way we can navigate and build this user flow? A lot of times --

CHARLES: And if the answer is no, it's so painful, right?

ROBERT: Yeah, exactly, so all that work that was put into building that carousel and all the components that built that carousel together is thrown out because it's just not a pattern that should be used there. Doing after the fact is really hard and really expensive and usually ends up in refactoring anyways.

CHARLES: I just want to point that out because a lot of people find themselves in that situation, where they're staring down at a pretty big cost. Now, the reasons why your app may not be accessible or many and good and you shouldn't feel bad, if you're finally in that situation.

ROBERT: I gave a talk at Nodevember a couple of years ago called Accessibility Debt and it's just like any technical debt. Your apps going to have it. It's okay. Don't get beat up about it, especially if anybody is trying to beat you up over it, don't listen to them. It really is just like any other kind of technical form of technical debt. It's something that you'll have to deal with and it is just something you have to work through. It's not world ending. It's just another problem to work through.

What are some things like everybody usually talks about like the things you can do, the basics for making your app more accessible like using all its attributes and instead of doing a div with an on-click handler, just use a button or don't overuse ARIA attributes. Are there any other things that I missed there like the basics of accessibility?

WIL: The biggest is the HTML structure. Screen readers and other assistive tech were built with the standards in mind, so if you're doing nonstandard things like putting divs in H1 and adding ARIA attributes within there, you're not going to have a great time.

ROBERT: Right or splattering ARIA roles all over the place, probably not a good idea. That will be harder to debug. Also fun fact, ARIA roles, while you can implement directly to the spec, you may still have bugs across the different screen reader combinations or assistive tech combinations, so that's fine.

CHARLES: Keep it super simple is what I'm hearing, like use semantic markup. If you're going to introduce a custom button, still make sure that it's a button.

ROBERT: Use the platform.

CHARLES: Yeah, use the platform. Don't fight the platform. Probably the best example of that is people implementing their own select boxes. That's the classic example.

ROBERT: Wil and I, our lives has centered around that for a little while. It's so true. Usually, it's the first thing people go to grab to reimplement because selects are just ugly. I think Firefox has the ability for you to style select options now like you can change the color and the font but you can't style that. The pop up that comes, usually that's the system dialogue, which a lot of designers don't really like. That's usually the first thing that people go to implement and that's usually actually the first thing that stop somebody from signing up. A lot of sign up forms that I see, if your date of birth is in a select format, that probably will hinder somebody that uses assistive tech from signing up.

CHARLES: Yeah. You just basically bounced that entire person. The thing is people don't appreciate the cost. This gets into the whole concept of accessibility that is how much money would that person or those group of people have actually brought into your site versus the cost that you spent redoing that select box. You might be thinking, "It only took this developer actually two weeks," but when you actually look at the actual cost over the course of your application, you're not factoring that into the decision to go with custom select box. Just in our experience, it's just the truly low cost option per quality of experience, that tradeoff there is almost invariably going with platforms select, right?

ROBERT: Yeah. Your secret sauce and the best UX that you're going to provide is not going to be nicely styled select box and seriously, a battle that I had to fight a lot was if you really want to implement a custom widget, decide if this is what you want to spend your time on because custom widgets aren't just quick and easy things you implement. You're not going to implement the select that's fully accessible across all the AT combos in a couple of days.

CHARLES: It's a lot of work.

ROBERT: Yeah. You're going to fall down on a huge rabbit hole.

CHARLES: Yeah. It is just you're committing to that work over the lifetime of your application.

ROBERT: Exactly, if you can maintain that now. If you implement custom check boxes and custom radios and custom input that of content edible for some reason, I think --

WIL: I think what I see is like custom date pickers --

ROBERT: Oh, I just had a rant about that.

CHARLES: Date pickers are hard and there's not really a good option. You just have to open your eyes to the true cost.

ROBERT: Right, exactly. That's what I always try to explain, just like you have ownership over this now and you now have to maintain this and you can't regress.

CHARLES: Right and if you do regress, it's your neck that gets choked.

ROBERT: Yes. A good way to put it. We've talked a lot about things that are just general accessibility but nothing specific to single page apps. I do want to say like a lot of other things that people recommend is like if you're using React, use like the JSX ES1 plugin to help you analyze if you're writing any JSX that might not be accessible or use like HTML_CodeSniffer or aXe to statically analyze the DOM that you have.

Those tools are great. I'd see a lot of people [inaudible] those things as like, "Look, this automated checkers says I'm inaccessible so I'm inaccessible," and that's not the case, especially in a single page apps. You can have 100% passing automated checkers but your app also could be 100% broken and why is that?

WIL: I don't know.

ROBERT: I'm wobbling the router question here.

WIL: Yes. I guess that would just be due to routing. In single page applications, they handle their own routing, whereas static web sites and whatnot, the routing is handled despite URLs and the browser and reloading pages.

ROBERT: Right. There are probably other major differences between single page apps but the biggest thing is the routing is managed by the client, the JavaScript and on a server-rendered page, you click a link, it's going to rerender the entire page for that next page and the screen reader and the browser know exactly what to do with that, to put the focus back to the top of the page and start working through the page for you. But with single page apps, you're just replacing a piece of the DOM and the screen reader has no idea of what's going on.

An automated checker cannot check for this. They cannot tell you if your routes are accessible. The reason I'm bringing this up is because this has never talked about. I don't see in accessibility talks and this is the thing that actually is most broken and makes your app pretty much unusable to anybody that's using assistive tech. There's ways that people if they're savvy can navigate around it but if they don't know what's going on, they're going to think, "All right, I pressed this button and nothing happened, so I'm just going to leave now because it's not working."

CHARLES: There's a great video that you point to me, I believe a guy from the UK who has recorded a bunch of his experiences using websites and apps --

ROBERT: Yes, I want to dig it up and put it in the show notes.

CHARLES: Yeah. Those are great to watch because you'll really get it.

ROBERT: Yeah and that's a really unique case because he's really savvy. I believe, I could be wrong but I think he might have said that he definitely work in tech somehow --

CHARLES: Right. He knows the workarounds and he knows these things but in some of the cases it's like, "If I didn't work in tech, there's no way that I would be able to use this website."

ROBERT: Yeah. This is kind of the crux of like if you ever listen to anybody that is an accessibility consultant, they'll say, "You will never ever be able to automate accessibility," and this why I tie accessibility so much to user experience, would you ever have a user experience test that can tell you in a binary fashion? Yes or no, that your app has a great user experience? No, because it's pretty subjective.

CHARLES: Of how does your users feel? right?


CHARLES: You can rate, "My users feel great."

ROBERT: Accessibility is like that because there's a lot of context that you have to carry around. It's all about context. When I transition from this page, the next thing does the user have enough context of where they're coming from to where they're going to be able to operate on that page. Is there enough information to achieve the task they want? That is pretty much the crux of why there is no binary yes or no for that because it's contextual. It varies from person to person but you do your best to make sure that that works and that you provide enough information to do something.

That's kind of a single page app as a crux. This is why we have a philosophy of testing as a whole. We don't test components individually because again, you can make all of your components individually accessible there but as a whole, they might not work together because you're not providing enough context on an entire page. We did this in one of the apps that we work on, which is open source, so we can link to the PR that Wil wrote for this and I wrote an entire routing documentation around our philosophy and the different things that we tried. How did we manage the focus in that application? I'm kind of just going to lob it over to you Wil since you did the work. Do you want to give context of the holdings and how that all kind of came together?

WIL: Yes. One of the features of the app is these panels. It's like this three panel system. When you click an item in a list, a third panel pops up and this goes back to the context thing where if you're using a keyboard, you can't see the screen. You click an item in a list, how do you know that third panel popped up? The solution isn't for a component. The component can't be responsible for this. The list can't focus the item. It opens or vice versa, so this is definitely an application concern, where we needed to check against the route and see whether or not, the pane is opening or was already open or wasn't open before and when this pane opens for the first time, it will just focus it and that gives a lot of the context that the user needs when they click it, like they click an item in the list, it focuses this third pane and they're on a third pane.

ROBERT: Right. We didn't even just focus on the entire div. We focused on the heading of the thing that you selected.

WIL: Yeah because focusing the div, it might read something off but not all the time. The main thing we're focusing on that third pane opens is the heading to let them know that the item you click, you are now on that page and reading that heading. This is the same thing that would happen if you loaded up that page statically and the screen readers would usually just focus on that first heading.

ROBERT: Right. That helped a lot. For a little bit more context, the middle pane is like kind of a master detail thing going on here and in the middle pane there that we have, it's an infinite scrolling list. You have different things there and one of them is like a package. If you select the package and without the focus management and focusing on that heading, you would have to go through every single package that's in that list which could be a thousand of them before you actually get over to the pane that you just opened because source order. You have to go through each one of those.

WIL: And it's the same thing is true for power users, not just screen readers. It's like if they want to use tab once that pane open, they have to tab through the entire list.

ROBERT: Exactly. It wasn't keyboard accessible and it made it really hard to navigate around with a keyboard because the focus just wasn't being managed. There was a lot of work that we did there. I want to focus on the routing situation there because if you can't navigate around with a keyboard in your single page app, like you click a link and it's not selecting the next thing that should be focused and you provide the right amount of context, your app probably won't be usable without a lot of trial and error to a user, which depending on what your product is, they may not have a lot of patience for trial and error, right?

WIL: Yeah.

ROBERT: It's really important to try and nail down the routing situation and there are some frameworks and things out there that can do it. In the app that we're talking about, it's React app and we use React router but we don't actually really hook into the router to handle that and that's because React router doesn't provide very much information.

WIL: Yeah. You can think of React router as more of an outlet system, where your routes can render anything, anywhere on the page. That's kind of dangers of accessibility and that's kind of the reason that React router can't handle accessible things very well because at any point, the route can just change the button on the page to look like something else.

CHARLES: Yeah. It just pop in and pop out. There's no deeper model, right? There's not --

WIL: Yeah, there's no tree.

CHARLES: Right, there's the internal state --

WIL: Like a component tree, yeah.

CHARLES: The internal state of what is happening is completely opaque. You can only analyze the second and third order effects of the React tree.

ROBERT: Right and especially if you have nested route components, it's really hard to determine. One of the things that I've seen people do is focus on mount, which is a very naive approach because what if you have three nested routes, they're all going to focus on mount and the last one that mounts wins and that's a focus for which nobody wins.

CHARLES: You all tried that, right?

ROBERT: Yes. That's part of the document that I wrote up. We tried three different approaches and we ended up landing on something because we were using React router that was more of like an application state thing, so we were checking props because we knew what the user flow was. We knew what the user, when they come to this page is trying to accomplish, so we're able to kind of figure out from where they came from or where they're going through props and focus the right things for them.

WIL: One of the examples, like we talked about the third pane opening and focusing the heading inside but what you know what happens when you close that third pane, you kind of lose contexts again, so we have to focus the previous list item that was active because if you focus back at the top of the list, they've essentially lost their place and the list of results. In that case, we couldn't use on mount. That list item is already mounted. We have to listen to props and we have to look at the route through these props to determine if that third pane was open. If it just closed and if an item in the list is active, it should have focus. It's a lot of logic going on. You have to really understand your app in order to make it a very good accessible app. You can't just sprinkle in ARIA attributes and focus on mount everywhere and think it'll work fine because you're accessible. You have to really know the flow.

CHARLES: Right. All those signposts point towards having a deeper application state, a deeper understanding of your application than just the render tree. At that point, it's too late and so by that, I'm definitely lobbying a Reach/Ember router.

WIL: Yes. We talked a lot about the React router and we can't really be too accessible with it but to create a React router to go out and there is now Reach router. Robert, have you heard any good things about that? You're the accessibility expert here.

ROBERT: I haven't played with it myself, so [audio glitch] things that were lost from the React router three to four transition, I think and also, it's actually accessible routing which is nice because it comes out of the box and you know how to implement it. I haven't played with it. I know Gatsby has implemented that for their V2, that's their default router now, which makes me really happy because a lot of static sites that were being built with Gatsby were very inaccessible and broken which made me sad but now, they're not with V2.

I haven't played with Reach router but one of the things that I think it provides which was missing from React router was transition hooks. It has a concept of like where you're going from and to for your routes, which really helps figure out what you need to focus. The one thing I will say about Reach router and I'm sure it's got to be configurable somewhere but I haven't really looked and the demos that I saw, they were just focusing the div of the content that's being rendered, like it kind of just wraps with a generic div or the tabindex="-1" and then focuses it. If you have a lot of content that's inside of that div, it probably will be confusing but at least, it manages the focus somewhere.

If you're now off in a no man's land, you have no idea what's going on, at least it focuses that. But if you want to really nail the experience to be better, I would recommend trying to figure out what you should focus inside of that route that you just transition to, what is the best thing. That's for React. There are other things out there for other frameworks, like I am on Ember accessibility team that's out there and we have Ember A11y, which just provide you a focusing outlet for you to be able to just drop in your app and then when the route changes, it does the same thing. It focuses the wrapping div of that outlet that was just rendered.

I want to emphasize more in talking about e-holdings work that we did that we were focusing the heading because that told you exactly where you're at and what package you're on or what thing you're on. You know the name of it and now you know how you can go and navigate through that.

CHARLES: But it's conceivable, so how would you do that with the Ember router? Would you just introduce some way to delegate down to a particular component? Like when I'm rendered into an outlet, it focus on this component?

ROBERT: That would be interesting. I actually don't know. It's been a long time since I've messed around with the Ember router and Ember accessibility. It's definitely a great first step and that's where it kind of came from. I think there needs probably some work done to help implement on what you want to focus. That's kind of where I was going with when we were first exploring the React router stuff and e-holdings was I wanted to have this like high order component that knew of your routing tree and it knew where you're coming from or where you're going.

Then from there, it would just tell you, this is the route that we're going to and there's need to be focus management done, like if this prop exists, then you can pick inside of that component that what you want to focus. It leaves it up to the implementor of what they need to focus but it could be probably just like a fallback to the general div because that's not bad. It's a good first step. There needs to be a little bit of work there to get that done probably.

CHARLES: Yeah. But it is worth pointing out that by starting from a position of having an externalized application state via the route structure, in an Ember application, you're starting 95% of the way there. An Ember application, at any point, you know where you are and during your transition, you know where you're starting and where you're going to end up and you know when you leave and you know when you got there.

ROBERT: Yeah. Having an Ember route pivot handler there, like where you're pivoting from and to is just so nice and it kind of made it click together a lot easier.

CHARLES: Right. Reach router looks interesting but as I understand, it's still couched in React components and it feels to me like this is a problem that ought to be solved, that ought to be framework agnostic. Because if I use something like Reach or I use some other routing library for another framework, some other tool might come out that I want to use and I shouldn't be locked in --

ROBERT: Right like what if I really like Redux little router, then I have to make a choice between Redux state that I like or accessibility.

CHARLES: Exactly and that feels like a false dichotomy to me. What I would want to be looking for is a platform independent or a framework independent routing library that really just helps you represent your application state and the concrete state in which your application is in at any moment and then, also be able to represent the transitions between those states fully and completely. Then if you have that, you could embed that into any framework.

ROBERT: Right. That would be really nice to have. Just like pull it off the shelf and help you out there.

CHARLES: If anybody is listening who wants something to do for the next 18 months, no one will thank you for 18 months but you had to get on that. We'll give you lots of thanks then.

ROBERT: [inaudible] compare with you. That would be awesome.

CHARLES: I would love to be part of that.

ROBERT: Yeah, that would be awesome. To kind of tie back to other things, I don't know too much about Angular but I'm sure there are solutions out there to help with Angular. I know Marcy Sutton used to do a lot of work in the Angular world, so I'm sure there's something out there that helps with that. I just don't know off the top of my head right now.

I wrote a Medium 'think piece'... No, I wrote a little medium blog post about how all of single page out for routers are broken. I was pleasantly surprised by Vue. Vue brings its own router and their router has a concept of before each, so before each route transition you can run some code and that really helped with implementing the live, the announcer pattern where you use ARIA live to announce something but even beyond there, if you wanted to dig in there, you could probably figure out where you're transitioning from and to and give that next route some kind of attribute that says, "This needs to be focused. Figure out what you need to focus there and inside that route, you can focus wherever you want." I thought that was a really awesome.

That's kind of the crux of what I was missing from React router. I wanted the concept of knowing where I'm coming from or where I'm going and I would help with everything but unfortunately, that kind of doesn't exists because they're just components. For better or for worse, they're just components.

CHARLES: The world is so much more than just components.

ROBERT: Yeah, a little bit off topic, I think it's kind of funny how React kind of just shoved everything into the Vue layer, just to make it all a component.

CHARLES: Yeah, it's very easy.

ROBERT: Yeah, until it's not.

CHARLES: Yeah, exactly. It's easy but it's not simple.

ROBERT: That's a lot of talking about how routes need to be done and what you can kind of do to manage focus. It's really about managing the context and how you can provide the most contexts. For somebody to operate on that information, can they complete this thing? What can you do to make sure that you've done this properly? What steps can you take to make sure that you actually are accessible and that your routes work?

WIL: Manually testing those screen reader is probably the biggest thing, you know?

CHARLES: Yeah. I think the biggest thing is really watching someone who uses assistive tech on a regular basis use your application and then trying to use it yourself.

WIL: Yeah.

ROBERT: Right. Yeah, I'll definitely --

CHARLES: If you can get a little bit of this yourself but then it's kind of like someone who is never used a mouse before and trying to learn something new. What really helps is seeing someone who's good at it and see how their expectations are either being met or not being met.

ROBERT: Yeah, it's definitely the best way. If you can find somebody that actually relies on assistive tech, there's nothing that beats that kind of feedback. If they get to your app and they're really confused, I see some people that just dismiss it because they just don't understand. That is the best feedback you can actually get. If they don't understand what's going on, you have --

CHARLES: That's on you.

ROBERT: Yeah, that is going to be what happens for everybody that uses that. Well, maybe not everybody because everybody has different experiences but it's probably going to be a thing that pops up everywhere. But if you don't have access to people that are actually using assistive tech regularly and are pros at it, WebAIM provides really great tutorials for how to use a screen reader. If you're using a Mac, you have a screen reader built in and you can use that called VoiceOver. If you ever want to turn it on or turn it off, its command F5.

WIL: You might have to have that shortcut enabled, though.

ROBERT: Really? I'm pretty sure it's quite default.

WIL: I thought I had to enable mine but I could be wrong.

ROBERT: It's interesting.

CHARLES: Yeah. They got great tutorial too. It's like it notices the first time you turn it on, so it tries to help you navigate bullet lists and select boxes and input fields and check boxes and all kinds of good stuff.

ROBERT: Yeah. It gives you a little bit of a boot camp but WebAIM also helps with specific to website and stuff because one of the things that we ran into while working on the e-holdings project is they're transitioning from a native app to a web app and there were just things that you can do on a native app that you cannot do on a web app. Just keep that in mind also when you're testing, there's just things that will behave differently. Like you're not going to have a lot of crazy key shortcut commands like you're not going to press command F to get to the search box if your app has a prominent search box because that is going to clobber a bunch of other assistive tech key commands.

WebAIM is really a good help with giving you the tutorial of how to test a web app and many different screen readers, so they have it for JAWS, they have it for NVDA, they have it for VoiceOver iOS, they have it for TalkBack, which is the Android screen reader. There's a lot of really good resources there for you to start using as screen reader and test with your app.

I highly recommend using it with a screen off. You gain a lot of context by looking ahead of your cursor and --

CHARLES: Yeah. It's true.

ROBERT: -- The example that I usually give to people is like, "I worked on Visa Checkout. I went through that checkout flow pretty much every day for a year and eight months in, even then turning off the screen, I was still lost." Even though I knew what each screen was and what each component was there, I would find myself confused of like, what just happened and it's because sometimes, you'll get into something like you have a dropdown that sets the focus inside the dropdown and then the dropdown disappears from the DOM and your focuses in nowhere. You're like, "What is going? what am I doing. I don't even know where I'm at," and I turn the screen back on and I'm like, "Oh, now, I know what's going on."

CHARLES: Right. It really is a lot like interacting with your application as though you were interacting with Siri or Alexa but with a keyboard instead, instead of voice commands. That's an excellent point. We would understand if where would Amazon be if Alexa couldn't successfully navigate those situations. The counterpoint or even the flip side of that is if you model your web application in such a way that it can handle that type of serial interaction, instead of the highly parallel environment to being able to perceive huge amounts of information concurrently, like you can on a screen, if that were effectively serialized, that means you could represent your application through nothing but voice commands.

ROBERT: Yeah. Did you provide enough context for me to build this mental map is really what I'm going on to?


ROBERT: Which I always thought was really interesting because I always wanted to know how my mom visualizes what a website looks like because it's wildly different than any of us. She's never been able to see what a website looks like. Does it look like a bunch of nodes and graphs and webs connecting to each other and how things pieced together? It's just a different way. But it's not only about screen readers, right? You can use a keyboard to navigate and that's definitely what we did with e-holdings is like can we tab through this [audio glitch] list, hit enter and go into the detail record, edit it, close it and go back and edit another one. Is that something that we can do with just a keyboard, not even a screen reader?

WIL: And with the keyboard, we're not talking about shortcuts to hit edit. We're talking about like tabbing over and hitting enter like people with accessibility issues would have to do.

ROBERT: Right and that's kind of a good segue into creating use cases. If you want to know if your app actually works, if your screen reader users or if your keyword users or if your dictation users are going to be able to navigate this app, create use cases. Things like actual user flows like how would somebody actually going to use this app? What task are they going want to complete? In that case, e-holding was like an electronic holdings management system for libraries. They probably want to get in there, add a couple of books or whatever you might have to their library and get out.

A use case could be like, "Can I search for this thing? I'm going to search for something specific. I'm going to go through the list, find the thing that I want. I'm going to add it, close the pane, go back and then remove one thing and can they complete that flow successfully without running into any issues or any blockers or any showstoppers." I can tell you before we did the routing management stuff, you would hit search and that was it. There was nothing else that you could do.

CHARLES: Yeah. They wouldn't even announce that anything can happen.

ROBERT: Exactly.

WIL: Yeah, and with these librarians, it's not necessarily a matter of they can't see the screen. It's just that they don't use the mouse because they're power users.

ROBERT: Right. They don't have any disabilities but that was essential to the workflow.

CHARLES: Yeah, exactly. Do that change. You just lowered the activation energy of that workflow by... What? three orders of magnitude?

WIL: Yeah, at least.

ROBERT: Let me click here right now. I can tab, now I can tab. Right now, let me click here and now I can tab, I can tab, I can tab. It's not as nice as just being able to completely do it through a keyboard. Through us making it super keyboard accessible, that also became super screen reader accessible and the people who use dictation were able to work through and get through the app and use it now, which was really cool. It really helps when you go for those things and create use cases to really figure out how a user is actually going to work through this app. That's the best way. Just get right through it. With Visa Checkout, we're like, "Can somebody buy something?" or if they don't have an account, can they sign up and buy? those were some of the use cases that we had because it turns out, those are actually pretty important to the business.

That all have been said, you also can test your components for accessibility individually because even at a smaller level, some of your components might have to manage focus. The best one I can think of is models. When you open a model, you should trap the focus inside of that model and you should be able to hit escape to close the model and when you close the model, it should go back to what triggered the model to be opened. These are all things that are individual that can be tested also.

But just because your components are individually accessible, it does not mean your application as a whole was accessible either. I don't want to paint the picture like, you don't have to care about your components accessibility individually because you do. It really does help but I think a lot of people miss the whole of the application, rather than individual with components.

CHARLES: Right. The components are the individual stitches but you have to follow the thread throughout the entire garment.

ROBERT: Right, exactly.

CHARLES: Man, what I really want to do is I want to find out how your mom visualizes website navigation and use that as a visualization technique.

ROBERT: That might be a fun webcast or something.

CHARLES: Yeah and then see like could we actually use it as a tool because I have to imagine, it's probably pretty simple. It's much simpler than what we think of when we think of a website because it has to be really condensed down to its essence.

ROBERT: Right. Yeah and [inaudible] users, they're not dumb. They have different ways. They know the gotchas. They know things that happen. They know there are different ways of getting trolled like my mom knows about focus jumping and she gets irritated what happens but she knows generally of what to do and it depends on her patience level. Like if it's focused jumping, she's like, "I don't need to use this thing. See you," and it's just not worth the frustration but there's different ways to navigate around like if you're a real power user, you might be able to recognize like the routing is inaccessible and you can navigate by headings or by regions or different landmarks. There's many different ways for users to navigate but to use those different navigation methods, you need to have a real strong coherent document structure. Your H1s have to actually be H1s and you have to have some things that they can navigate around to kind of work around those things.

It's interesting and basically, do everything you can to help those situations. If you can provide semantic markup and give it a proper structure. If you can do the focus management, it's going to help everybody. It really will. I did see when we went through the use cases and defining those things, we actually learned a little bit more about our product because we had to put ourselves in a different seat and think about it because you get real close to it. You go through that same flow nine million times and you pick up real power user things that you can do like, "I'll work on that. I got this. I'll click this button. All right, we're good." Somebody that's going through it for the first time and you put yourself in that seat, it kind of opens your eyes a little bit and makes the experience better for everyone. I think that's kind of the underlying tone there. That's the message.

WIL: Yeah, accessibility makes things better for everybody.

ROBERT: That was a lot of content thrown at you there. We covered what is accessibility and why you'd want to do that and then kind of like more the basics things and how automated tools are really helpful and they can help you pick out things like using improper roles and nested things but they're not going to be able to tell you if your application is truly accessible or not and never will, unless we get something like a headless screen reader, where you can write automated tests for in that fashion but you're not to get something that will just run over you app and go, "Yup, you're good."

CHARLES: Even so, it's a matter of user experience and that's not something you can get a thumbs up or a thumbs down to. When you as a user, it comes with application, you know when you see it but I would say until we have androids that accurately simulate human beings, I don't think we're going to actually have automated testing.

ROBERT: Yeah. There's a joke for accessibility consultants. It's like if you put four accessibility consultants in a room and tell them to give you an alt attribute for an image, you'll get four different alts.

We talked about the automated checkers, right? They'll not going to get everything for you and we talked about single page apps specifically in the routing and how we handled the routing in a React app and then how you can probably do it in an Angular app and how you can do it in an Ember app and Vue and different methods of how you can kind of attack that. We're definitely giving the link to the document that I wrote and the PR so you kind of see the real 'how' of how we did it because that would probably be helpful.

I think there was a lot of good information there, so I would like to thank both Charles and Wil for being awesome co-host on this and --

WIL: Thanks for being an awesome host.

CHARLES: Yeah, thanks for being an awesome host. I should say, you're welcome.

ROBERT: I tried, I tried. We are The Frontside. We do accessibility consulting and training. If that's anything that your team needs help with, we're more than happy to jump on a call with you to kind of figure out what your needs are and what you need to do. If you need a WCAG support statement, if you need to audit, if you just need to figure out what's the next steps for you to even do, we're more than happy to help you sort through that. To reach out for that, you can contact us at Contact@Frontside.io or Sales@Frontside.io or Info@Frontside.io. You can contact us at any different ways and we'll be more than happy to help you.

As always, thank you Mandy for producing the podcast. You're awesome and the next episode, what we're going to have is also still accessibility related and I'm really excited about this. It's about writing a proposal to the Unicode Committee and getting it accepted so basically, writing a proposal to get an emoji added and that's with Amberley and Amanda. They wrote three or four Unicode specs and actually got them accepted for, I believe it was sign language and deafness. That's really cool and I'm super excited for that because they'd be the first people that I've ever talked to that have actually created an emoji and gotten it accepted.

WIL: Yeah, it's pretty cool.

CHARLES: Yeah, that must feel great.

ROBERT: Yeah, it's going to be awesome. That's our next episode. If you have any ideas or comments or anything, you can tweet us at @TheFrontside on Twitter or you can contact us through any of the emails that I talked about earlier. We're always open to hearing feedback. Thanks for listening. Take it easy, everyone.

Listen to our podcast:

Listen on Apple Podcasts