I just visited flutterfolio.com. The very first thing I tried to do is enter a username and Tab to the password field. But Tab just catapulted my input focus to the browser address bar. I mean, come on.
The accessibility is a bigger concern, but flutterfolio.com also
A) lags on my work machine when tabbing through fields
B) doesn't handle HDPI screens well (the entire interface looks blurry)
C) keeps fields visually indicated as selected even when I click outside of the browser (if I start typing, is it going to go into the field or not?)
D) has a separate touch mode? Touch doesn't just work?
Looked up another of their sample apps:
A) doesn't respect my OS scroll direction
B) lags on scroll
C) mousing over any of the buttons has a 0.5 second delay before my cursor turns into a hand icon.
And all of these sample apps take about a second to fully load for me, despite the fact that most of them aren't doing anything complicated.
The "we can rebuild the browser in Canvas" idea is alive and well in GUI toolkits, but I have never seen an implementation that didn't have these kinds of basic problems. How many UI toolkits do we need to see that can't figure out how to do scrolling before we acknowledge that having one team try to independently reimplement decades of work on the web might just be a bad idea?
I really appreciate the feedback. We clearly have more work to do on the Web side of Flutter. Unlike the Mobile side which has shipped 100,000s of apps, the Web side may be up to like 1000. :) So many more issues to address as we work closely with more users to get their apps into production.
If you'd like to track progress on any of these, I'd encourage you (or anyone else reading) to please file issues at flutter.dev/support. We read (and try to address) each and every one which comes in via Github.
The CanvasKit renderer[1] is highly concerning to the general health of the web. It's nice that it's easy for developers, but it remove all the power of the web.
To an end user, their accessibiliy tools, & user extensions, the page might as well be VNC'ed in to some system. Everything is opaque. There's no HTML elements.
I can not state strongly enough how immoral & unethical this path is. Bending the web to the developers will & breaking all the contracts of what a web page is a grevious act, is enormously harmful. This is one of the worst possible things that could happen to the web. Please for god sake don't do this to us. Don't injure the web like this.
This will be raised again and again and again until the ends of time as an incredible black mark on the name of Flutter.
Edit: constructive feedback welcome. Whether it's me or others, folks pointing out this peril often find themselves hit by downvotes as I do now. This seems like an existential challenge for the web, to keep humans able to be in the loop when browsing the internet. We would all love if you would defend or explain or have something to offer in your critiques about this crisis.
I appreciate that building something that works even this well targeting Canvas must have been a metric ton of work, and I'm impressed. But that's kind of the issue -- it cannot possibly be more work to maintain a separate rendering/compilation pipeline for browsers than it is to rebuild the entirety of Chrome's DOM engine in WebGL for CanvasKit.
I just don't see how Flutter is going to keep pace with "native" web apps when the team is constantly playing catch up every time a new medium, input method, or browser feature comes up.
Handling scroll is hard enough, is this going to handle WebVR? Can I control this web app with a voice assistant?
Sorry if this is something of a dismissive answer, but I'm not sure what the right way is to file an issue that boils down to, "it seems like you picked literally the hardest possible way to build a web GUI toolkit, and I'm very impressed that you have anything that's usable at all, but I also don't understand why you did this."
I don't know, I don't want to be overly negative here, but I don't have any confidence at all about the overall strategy that Flutter is pursuing. In order to get accessibility to work, you're basically going to end up building a DOM representation of these apps anyway. It's just now you have to build a DOM representation and you have to completely reimplement the rendering stack and capture all of my events? How are you going to make something like that performant across every browser?
I want to echo a few other comments I've seen here that this result isn't really surprising to me. I guessed before I clicked on the demos that they probably wouldn't handle scroll events well, because I know how you're building Flutter for the web, and I have never seen any GUI frameworks targeting pure Canvas on the web that don't have basic errors like this. If Flutter ever managed to be the exception, that would be very surprising to me. I don't know what Flutter's internal process looks like, I'm not going to tell you you're building something wrong. I only know what the end result looks like, and I know that end result is not going to radically improve any time soon.
It's not my place to tell any team what to do, and nobody is obligated to care what I think, but... I don't know, the demos online do not leave me feeling confident about Flutter's future. I don't think the experience as it exists today is the result of a bunch of isolated issues/bugs that can be fixed one at a time.
Its nice to have absolute control of the rendering going over canvas if you are a flutter platform developer and i think even playing the catch-up game they should get there eventually.
But i don't think this is the real issue here.. Flutter is more of a Flash and GWT lineage, and that's fine if that's what you are looking for.
But i bet on the Web every time, its the most popular UI (and now much more) platform ever. So while some will go through the Flutter and React way of doing things, most will just stick with the web.
The web is the most sophisticated platform ever built. It gives you a lot of power and freedom, so i don't understand why some prefer to get stuck into a cubicle when you have so much space?
Its nice to have frameworks to make things easy and developers more productive, but the ones that do this without alienating the developer from the bigger, broad and sophisticated framework are just better.
Just let them be the new Flash, now at least open and over the web, and look for other approaches if you want to maintain real control and freedom, instead of delegating to the platform developers.
The price of infinite freedom in UI toolkits is that you get nothing for free. This has consistently been untenable: you quickly get stuck in the weeds reimplementing basic behaviors.
Supporting new and unique APIs/inputs/etc like WebVR is a pretty classic cross-platform problem and it's usually addressed by allowing native hooks. Flutter has MethodChannels which let you call native methods in the underlying "platform host". It also supports "PlatformViews" which let you embed native widgets.
It doesn't really need to have instant parity with "native" web apps at everything. Maybe Flutter is a poor fit if you have extensive native requirements. That's okay.
It's not just about building special experiences. I want to be able to browse normal websites in VR -- VR and uncommon user agents introduce a new set of user inputs that Flutter will need to handle if it doesn't want its websites to just break if someone loads them up in a headset.
Currently as a web dev, if I want my website to be usable by a voice assistant, I don't have to do anything. If I want my website to be usable in VR, I don't have to think about it -- that's Chrome's problem, not mine, I just expose the exact same controls I've always used. If someone is building a user agent that interacts with my site in a novel way, as long as I've built my site well everything just kind of works, because of how the web is architectured.
But if you're rebuilding all of the controls from scratch, suddenly you can't think that way anymore. You have to think about the controls and gestures that every medium exposes, or else you have a website that just randomly breaks depending on what OS/platform a user accesses it from.
> and it's usually addressed by allowing native hooks
Not on the web though. I mean, for obvious reasons on the web we can't just expose native OS methods for websites to call. Flutter doesn't have a choice on that front, Google, Mozilla, and Apple realistically aren't going to break browser sandboxing just so the devs have an easier time.
It's true of course that as a cross-platform toolkit independently built on top of another platform, Flutter will have to consider new controls, gestures, and changes in many platforms. You could say the exact same things comparing Android/iOS/Windows to the cross-platform web: it's very hard to do and there are plenty of compromises, but it's still very valuable. There's been far more handwringing over the web's deficiencies as compared to native OS platforms; why reinvent the wheel with some crappy dynamic language called JS, why rebuild everything with a custom style language and layout engine? Yet Electron and web apps continue to dominate because the trade-offs can still be worth it.
The web is not really adopting hordes of new input modes like WebXR every year, there'll be plenty of time for Flutter to consider new stuff like that before it approaches relevancy in mainstream product development. Accessibility features like voice assistant support don't necessarily come free either. Rich web applications usually don't produce simple/accessible DOM structures even if you're developing with web-native tooling. Flutter or not, accessibility practically always requires some level of manual annotation in web apps, which is Flutter's primary use case.
> Not on the web though. I mean, for obvious reasons on the web we can't just expose native OS methods for websites to call.
I meant native-to-the-web. Flutter's MethodChannels & plugin system works on the web too. If there are browser APIs that you need to use, you can setup hooks to call them from within Flutter.
It's disappointing the Flutter team members haven't addressed this justifiable concern about their big-picture strategy. I would really like to hear what they have to say on this.
On mobile, you can hook to low level native input events.
On a web browser, <input> is all you have and because of the abstraction you cannot build it from the scratch using lower events. Think iPhone text selection.
Flutter on web may cater a specific category of graphical intensive apps, but not all apps because <canvas>. It will fall to the same uncanny valley where Java applets where back in 00s.
Multiple countries around the world have legal standards around accessibility. Then you have the ethical, DE&I, and business reasons to enable disabled users equal access to the web platform and apps through screen readers and other assistive technologies, i.e., accessibility.
It’s certainly not a uniquely American concern.
I assume that you just haven't run into this specific term before, but if that's not the case then comments like these are why the web community overall should oppose web technologies that leave accessibility decisions solely up to end developers or that don't make accessible options the default.
The web works as well as it does because (in general) it's accessible by default. There are lots of open problems on the web regarding accessibility, but we should be moving towards being more accessible by default, not less -- because if we want a universally accessible app platform, developers and corporations can not be trusted to make these decisions on their own, and laws will only affect specific industries and countries.
That's part of why Flutter devs talking about dev decisions that can increase accessibility ring kind of hollow to me. Accessibility has to be something that's pushed at a technological and cultural level if we want the web to be a platform that leads the way forward on these issues.
Currently I'm counting four or maybe five different HN users in this thread indicating this affiliation. I appreciate this open look into the organization. We're probably getting a more honest idea about what really goes into the sausage. Still, your PM might be less happy about this thread...
I happen to be on a 1.5 Mbps connection at the moment, and "about a second" sounds wonderful. The demo apps I opened took closer to 30-60 seconds to load.
Strange how I don't see many of your issues on my side.
(parent issue) Tab tabs through text fields only.
A) No lag on my laptop
B) Zooming to 400% reveals pure vectors for text as well as control decorations, even the fingerprint icon. Only the logo becomes pixelated but that one could of course simply be an SVG instead and not Flutter's problem.
C) Yes, this is a common issue with web apps. One often needs to look at web browser focus. I'm unsure if an event is even sent to the page as the page itself loses focus.
D) I think it does and this is just a tech demo for adaptive interfaces if you want them.
Could be the browser/platform you're using? I'm not sure. I've seen some reports here that the demos work better in Chrome, but that's not exactly a point in Flutter's favor; this is supposed to be a cross-platform toolkit after all. Regardless, odd.
A) I'm on Firefox on a Mac Book Pro. My feeling is that should be able to handle a login form, but I mildly suspect that Flutter's dev team is not testing on Firefox at all.
B) Here's what I see when zoomed to 400%. No vectors anywhere, not even for text: https://i.imgur.com/7QFC40v.png Nothing that I see on the page is being handled as an SVG. Very strange to me that you're seeing vector graphics, as that would imply that the form isn't being rendered onto a Canvas for you?
C) This shouldn't be a problem with web apps that use :focus/:active CSS selectors. I understand if a from-scratch webapp doesn't handle focus loss, but I would expect any GUI toolkit to be able to.
D) On the web touch already for the most part just works. There's nothing on this login page that should require a separate interface. I think it's a problem if Flutter only provides adaptive interfaces "if you want them"; in a good GUI framework you should be getting the stuff that the web already handles for free regardless of whether or not you "opt in" to having your site work on an iPad.
I think you're missing the point of touch mode: mouse and touch actually work fine in both modes. The touch mode is just there to demonstrate how Flutter adapts the layout depending on the target modality: see how the controls get larger when you select touch mode.
The Medium post[1] about Web support explains some adaptations they made for Desktop browser, including:
We also increased the default content density, because mouse pointers support a tighter density than touch devices.
I expect that in a "real" app the default mode is adapted automatically based on the device type.
B.2) I bet that it doesn’t respect a wheel multiplier either. It is 3 down keys by default, but I set it to 6 (7 at summer). Even VSCode annoyingly does not, but at least it has its own setting for that, which still only affects text buffers and not everything else.
I know I do almost nothing on HN but complain about accessibility, but it's really important to me because, well, that's how I uze my tech. And so far, sadly, Flutter hasn't impressed either with its previous versions. Edit boxes in particular were a major pain point, and things like you're describing were also less than optimal. I never knew where the focus was and what it was doing. With broken markup there was at least ways around it. But I'm scared of things that render on the canvas. Most browsers actually get this right and have support for these things that don't require ugly hacks if you use the DOM. And there is no real way to interface with assistive tech from JavaScript except through it. So you have to somehow still actually represent your apps UI in the Dom in some way.
To be fair there were some flash sites I’ve still never seen anything that equals. I think the way flash marketed to more designers and content creators was both what made it awesome and maybe its downfall.
It is the same. But at least runtime, WebAssembly and canvas are open source and natively integrated to the browser. Less security holes, but the same UI integration problems.
Yup, that happened to me. Then I clicked again and tab worked correctly.
It will be hard to overlook little details like this -- and there will be a million tiny details like this -- when decades of browser development have led us to expect a DOM experience that Just Works for UX like that.
If your situation is like mine, you have to wait for some
fifteen seconds or so. And then greeted with a login input.
Also, my CPUs were at around 50 % the whole time.
I'm sorry, but if that's the “future of web UIs”, I'd rather
stay with the “troglodytes”, who write simple HTML forms
with a bit of vanilla scripting here and there.
A) lags on my work machine when tabbing through fields
B) doesn't handle HDPI screens well (the entire interface looks blurry)
C) keeps fields visually indicated as selected even when I click outside of the browser (if I start typing, is it going to go into the field or not?)
D) has a separate touch mode? Touch doesn't just work?
Looked up another of their sample apps:
A) doesn't respect my OS scroll direction
B) lags on scroll
C) mousing over any of the buttons has a 0.5 second delay before my cursor turns into a hand icon.
And all of these sample apps take about a second to fully load for me, despite the fact that most of them aren't doing anything complicated.
The "we can rebuild the browser in Canvas" idea is alive and well in GUI toolkits, but I have never seen an implementation that didn't have these kinds of basic problems. How many UI toolkits do we need to see that can't figure out how to do scrolling before we acknowledge that having one team try to independently reimplement decades of work on the web might just be a bad idea?