When I led the Google team we had pretty good analytics. `flutter` was for a long time opt-out with analytics and would phone home to Google and report usage. We filtered out docker containers and things that looked like CI and saw usage near 1M monthly actives when I left Google a couple years ago. I'm sure it's up since then.
There are 10s of millions of web developers in the world. Doesn't shock me that Flutter could be over 1M monthly (and probably several million annually).
There are just a lot of apps and app devs out there.
That said, I literally only use Flutter as a hobbyist.
I made a basic web game for a friend, and a small web app that helps me manage my music lyric videos. I ultimately picked Flutter because I find it easier than actually building websites.
Dart is amazing.
But I'm not deploying to millions of users. At most 10 people have seen my Flutter apps.
It's VERY good for creating quick crud apps with Firebase.
Just counting the number of us who have it installed and might of spun up hello world says nothing about it's actual market share.
To be clear, I absolutely love Flutter, but it's still not something I really see job listings for.
on behalf of one of the Flutter founders, thank you for the thank yous.
I bet they'd mean a bit more if it wasn't coupled to a lengthy thinking-out-loud post that casts doubt on it, based on how many job listings you see, for something you don't look for job listings for.
> I bet they'd mean a bit more if it wasn't coupled to a lengthy thinking-out-loud post that casts doubt on it, based on how many job listings you see, for something you don't look for job listings for.
Did you really just thank the guy for saying thanks, but give him a big fuck you in the same breath for not being a 100% kool-aid drinker?
No, I sarcastically thanked him for free-association about the # of job listings he sees for job listings he's not looking for and what that means about the language.
I'm guess I'm capable of being curious and engaging in conversation, beyond sitting in the bleachers grading everyone on my scale that at 100% is "koolaid drinker.", depending on if I feel like the things they're talking about are on Team A or Team B I've identified.
Then again, that perspective would sort of come naturally if I were reading hysterics into every comment I read. "Big fuck you" is textbook catastrophizing.
My point was you just can't count SDK downloads as developers. Is every CI/CD pipeline a developer ? The founder of Flutter who posted that stat has a startup which is heavily invested in the perception of Flutter being popular.
In the Unity community we aren't afraid to critique Unity and even suggest trying other engines. Of course Unity isn't open source so you don't get articles like this where someone claims it's time to fork.
At the same time, Unity uses C#, so you can take your skills, and even some nuget packages( with a bit of work) elsewhere if you want to switch.
Flutter requires a higher time commitment since I have to learn a language that's exclusive to Flutter. If this ship goes down I can't use Dart with anything else.
Anyway, I think job postings are a much better metric of adoption. I like Flutter, I've used it for 4 years and I want it to succeed. I don't want to go back to react native!
If you think its not popular until that increases past one of: JS, Python, Java, C#, PHP, C/C++, Ruby, Go, or SQL...I don't really know what to say. :)
Flutter in Kotlin instead of Dart would have killed! Alas. Hopefully, Kotlin Multi-Platform holds up as good as Flutter has (the rendering architecture based on Skia seems similar between the two frameworks).
Just seems like personal preferences. I've used languages with extensive pattern matching and don't feel like I'm missing out without them. Don't really see the problem with Gradle or Static Delegation either.
Then you won’t understand what made Flutter “click”. I work with Kotlin, Gradle and Compose on a day to day job and it is absolutely dreadful experience. Sure, it beats working with Java… Java 8.
Yes, I agree, trying to use Kotlin instead of Dart would have killed Flutter.
I've been hearing about Kotlin multiplatform just as long as flutter, and writing Kotlin since 2019.
Kotlin is a horrible daily driver.
I worked on Android for Google at 7 years, and maybe it'd have a better chance if it didn't win internally. As it stands, there's too many organizational boundaries created, and each organization holds itself accountable only within itself, so it's sort of the worst of all worlds, impedance mismatches everywhere, and nobodys fault they exist
I'm always stunned to read wish casting about Kotlin because there's ~0 path to even basic things that change productivity dramatically, like hot reload.
> I don't hate Kotlin, it feels like Google's answer to Swift
You have the wrong order: JetBrains created Kotlin[1], and anything Google had to do with it was just adopting their language. Same story for Gradle choosing it over Groovy
> Can we at least agree it's weird Google is trying to promote 2 different languages for multiplatform development?
Google doesn’t promote anything. There are two competing teams with their own agendas. Flutter’s bills are mostly paid by internal usage, as far as I know. Android is mainly focused on Android part of Compose.
Multiplatform Compose, on the other hand, is mostly pushed by JetBrains to eat some of Flutter’s lunch and promote Kotlin usage to drive their IDE sales.
> Let me compare Unity to Flutter for a bit. When Unity jobs are hard to come by, I'm still an OK C# dev.
Flutter/Dart doesn't offer they same freedom. Say what you want about Microsoft, but C# can essentially do anything. Including keeping me employed.
I might be wrong, as I’ve worked with .Net professionally, but I doubt you can jump from Senior Unity C# developer to Senior Asp.Net developer. Language is a small part in modern development.
>I might be wrong, as I’ve worked with .Net professionally, but I doubt you can jump from Senior Unity C# developer to Senior Asp.Net developer. Language is a small part in modern development.
I'm not exactly senior to senior, but I hopped from mid level hobbyist Unity dev to professional Unity dev , to mid level .net dev. I have a very specific niche though.
I'm very comfortable with my career.
>Multiplatform Compose, on the other hand, is mostly pushed by JetBrains to eat some of Flutter’s lunch and promote Kotlin usage to drive their IDE sales.
Android Studio is free ? Are they really making that much money off users using Kotlin outside of Android Studio ?
Android Studio is subsidized by Google and works pretty much only Android. The moment you have to work something more, e.g. some Web stuff you need to pay for IntelliJ Ultimate as it’s only IDE that supports Kotlin.
They even have Ktor (Kotlin web framework) plugin available only in Ultimate. For any non-trivial Kotlin development you need to have Ultimate.
Only for the crowd, I need point out that no one's arguing there's more Dart jobs than C#.
Re: multiple frameworks, I worked at Google and would argue I know as much as anyone does exactly what happened there, so I can't agree it's weird, per se. Handwaving, I'd say that's because the situation seems more 'certain' or 'settled' to me, for good reasons, but ones not worth getting into.
There's absolutely tons of threads to unpack in your comment, I wish we were in person.
Speaking generally, based on observation you gravitated most towards discussing qualities of a job one might or might not get:
I essentially left my job at Google for no paycheck, partially because Google x Koyaanisqatsi, partially because I just couldn't imagine having to go back to write Kotlin and/or Java day to day and being criticized either way. That being said, it's deflating seeing Flutter job listings after getting into this whole industry from iOS dev. Thing with Flutter jobs is there's tons of low priced job seeking competition that are smart as hell.
That's because it is unusually effective in environments where people can't afford a Macbook only for iOS dev, and at this point in its lifecycle, it self selects for people who get experience with new frameworks
(On a serious note), inside Google you tend to get more easily promoted if you launch new products than if you maintain existing products, so it could be tempting for the employees themselves to follow in that direction.
yeah, it fascinates me - both Apple and Google started these new UI framework things after Flutter, and both didn't land it, but instead, landed some half-baked thing where individual components can render live in the IDE.
That may sound more useful than it is: think "oh I can specify a dummy title and subtitle for a list view cell in code, then open a special pane in the ide to preview a source file, and then resize the window to imitate different screen sizes!", not, oh there's a bug, fix, save, insta-reload, verified fix. No holistic screens, or loading info from a database, or mutating data.
I assume this is an extremely hard problem. I remember trying to get Obj-C hot reload working 15 years ago and it kinda half-worked some of the time. And its not like I can edit the C++ my Flutter app is linking against and get instareload.
But still, it's one of the things thats easy to point out and drives home that these once-a-decade-or-two frameworks started in response to React Native/Flutter aren't addressing the fundamentals of what made them a sea change. Industry is risk averse and we make too many decisions similar to staying on track to launch MacOS 9.110 in 2024, instead of biting the bullet to get OS X done.
> both Apple and Google started these new UI framework things after Flutter, and both didn't land it
For Flutter to have fully landed it, it would take more than just hot reload. I'm not familiar with the latest but what are the chances of the Flutter rendering stack "going native" (as in not drawing to a Skia canvas or similar) on at least Chrome or Android? Or, is that the wrong question?
Yeah hot reload's just a one-off example that I think is the simplest way to relate the experience of SwiftUI vs. Compose vs. Flutter, though, the first two are long in the rear view mirror, I've been Flutter only for about a year.
Good question -- I don't think its the wrong question? Maybe? :P Tough to phrase on my end too.
I only feel native speaking in iOS or Flutter, even with the Android experience, forgive me: I guess I'd summarize it as "yeah, you're right. if you're wondering if they're ex. calling CoreGraphics on iOS, no" (well, they do, but for text rendering. Not for a red rectangle)
Generally, Flutter Web is rendering into a WebGL surface using Skia -> WASM (2.2 MB download! modern miracle). macOS, iOS, (done) Android (soon) are switching from Skia to Impeller. I think the thrust of your question is "are they still bundling the render engine?" and the answer is yes.
I find the way you phrased it intriguing, in that, before reading your post I'd say drawing to a Skia canvas is as native as you can get. But I realize now that means I was overindexing on "close to the metal", and its apex not-native to say "gimme the framebuffer and i'll take it from there"
I'm a bit picky with design stuff and I loooooove that Skia's bundled and I can rely on it cross platform
https://flutter.dev/showcase has a few examples. Flutter is everywhere at this point. TikTok in China (Douyin), PUBG Mobile, MGM Resorts, eBay Motors, Kijiji.com, Grab are a few off the top of my head. And of course a whole bunch of Google apps (GPay, Earth, Classroom, etc.), Toyota cars, LG TVs, Google's own hardware devices, etc. Tonal is probably the app I used most often that's Flutter. Caribou Coffee, Betterment, Norwegian Cruise Lines, NuBank, Realtor.com are other random apps I've used that are Flutter.
Do you have a source for that claim? I'm pretty sure that is incorrect. Bytedance has said they use Flutter but for other projects.[0] Douyin is not mentioned anywhere.
I don't have access to any devices with Douyin on it, so I cannot confirm current usage.
My knowledge of their usage is from discussions with ByteDance some many years ago when I was in charge of the Flutter project at Google. At the time they were using Flutter in 50+ applications (probably most of them internal), including Douyin (TikTok for China) as well as physical hardware kiosks on their campus, web apps, etc.
I just tested Douyin 23.5 from 2022[0] and it does not have the two finger scroll bug.[1] It is of course possible they were carrying their own patch for it all this time but that seems unlikely. I doubt they are using Flutter or if they did it must have been a long time ago.
[1] https://github.com/flutter/flutter/issues/11884 - click on the search button in Douyin on the main page and search for something to get to a regular ScrollView. You don't need an account for this.
I wouldn't say its unlikely they carried a patch for it, I just wrote a framework patch that I apply at build time in CI and locally.
I refuse to sideload arbitrary APKs, especially from bytedance. I feel bad because that is irrational, you did, and it'd be really helpful if I did and just did this myself, but, you should install FlutterShark and check: https://play.google.com/store/apps/details?id=com.fluttersha...
Why is "bug" in scare quotes here? It was most definitely a bug.
>I wouldn't say its unlikely they carried a patch for it, I just wrote a framework patch that I apply at build time in CI and locally.
The actual fix was pretty involved. I doubt a large company like Bytedance would want to carry around extra patches at the gesture level that make the dev cycle more difficult. Having one person carry a patch on their local machine is a different story.
Anyway, the Bytedance blogpost says only 200 devs are using Flutter which would make no sense if it was used in Douyin, and LibChecker[0] returns no results for libflutter.so.
> Why is "bug" in scare quotes here? It was most definitely a bug.
Is it? I thought it was cool, I can't think of why its disruptive to scroll a list faster if you scroll with more fingers.
> I doubt a large company like Bytedance would want to carry around extra patches at the gesture level
I'm a solo endeavour, and I spent ~30 minutes to do exactly this (patch gesture behavior) two days ago. I was stunned how easy it is. But I grew up on versioned closed source dependencies on Apple iOS frameworks that you had to patch the runtime at runtime to fix, so I'm easily wowed.
> 200 devs are using Flutter which would make no sense if it was used in Douyin
Seems reductive: "Only" 200 fulltime, 800 in the company...and we're in a discussion about how 50 maintain _the entire framework_. :)
If you have an Android phone, get FlutterShark and check which of your installed apps use Flutter. It's a surprising amount actually. For example, I just recently discovered that Supercell uses Flutter for their Supercell-ID flow.
It was 8 out of 86 apps for me (9% of my apps). Including Hacki, the HN client. It's a good variety of apps too, from big billion dollar companies to tiny games I downloaded from F-Droid.
Since the beginning of Android apps could see what other apps were installed on your phone without asking for special permission.
They finally added a permission for it - QUERY_ALL_PACKAGES - in Android 11 (2020). Unfortunately it's one of those stupid permissions like filesystem access where you have to apply for it via a form, and only whitelisted use cases are allowed:
> Permitted uses involve apps that must discover any and all installed apps on the device, for awareness or interoperability purposes may have eligibility for the permission. Permitted uses include device search, antivirus apps, file managers and browsers.
I suppose that situation is slightly better than "anyone can do it for any reason".
If there would be 1M flutter devs per year and on average team had 10 devs (rather rare most mobile teams would be smaller except really big apps) and development of app would take ~1 year you would expect minimum 100k new flutter apps in app store every year. That seems highly unlikely.
I suspect there's a lot of larger apps out there which use Flutter for one feature or set of screens but the majority of the app isn't written in Flutter. You could have a team of 100 working on a app where only three of them actually touch flutter on a regular basis, but depending on how exactly you capture analytics all 100 of them may get counted.
I‘ve worked for over a year, almost full time, on an internal Flutter app that never saw the light of day (maybe later, if the stars align). Many apps just get their support section revamped with Flutter because there’s no need for that to look fancy; but they‘ll still count against that quota. Many devs prototype their side projects with Flutter, but never release them. Etc
I'm pretty sure many/most professional app developers are in long-term projects, working on the same app(s) year after year. So for those devs the app count never goes up. That's at least my experience -- development work only stops when the app is pulled off the stores for business reasons.
Yes. Realistically off by at least a magnitude. There is literally zero chance there's 1 million active developers working on this experimental Google Tech out there.
while I'd agree with you that 1 million is too high, Flutter certainly is not experimental Google tech. It's widely used in particular internationally. I talked to a ByteDance guy two or three years ago and I think they alone had about 1k Flutter engineers. Nubank, Alibaba, BMW, Ebay use it extensively.
Flawed metric but eyeballing VScode Extension installs, Python 140m, c# 30 mil, Go 15 mil, Flutter 10 mil. It is very, very popular.
Just looked via google, and while no one really knows, common estimates of the number of web developers seem higher than that, on the order of 27 million.
Your point may still be valid, but... is 3-5% really that high?
(No dog in this, I'm not doing web development at the moment, although I have and probably will again.)
It's possible that they also count students and amateurs such as me, who only used it in a small project never published. There could be a million I guess, but professional developers are probably a lot less.
I think this is a strong possibility. For example, I don't work in programing and don't intend to ever work in the field. I do it as a hobby. I downloaded flutter, played around with it for a few weekends and decided it wasn't for me. If you go solely by number of downloads I would be counted as a 'flutter developer'
There's 200k weekly downloads of node on npm. A million flutter devs means that there's a comparable amount of node devs to flutter. I'm not sure how else to estimate it, but there's 170k github stars of flutter compared to 100k node, and a roughly equal amount of forks.
Github shows 98% of code is "Shell" on top of that there are also many different package managers that people use: fnm, volta that are build with rust. You can also install nodejs using homebrew.
No. It downloads binaries from nodejs.org. There is no "official" canonical means of managing Node versions. It's just a grab bag of community tools like NVM.
My experience when I last looked into Flutter was that it was heavily used in India - at least based on who participates in the community online - and much less elsewhere. That's not a value judgement, just might be a factor in why you don't hear much about it if you're not in India.
they're just too close to the problem and overestimating their relevance.
just because someone cloned a repo once or an NPM package was downloaded doesn't mean someone is a flutter dev. I don't know what the best metric would be, but its relevance is definitely overstated
Yeah, people like to count their "downloads" and call that their "users", but it's at least an order of magnitude difference in my experience. Sometimes a 100x difference.
I don’t sub him on YouTube - but I’m not lying when I say everyone I know IRL who does sub his channel has worked as a barista at some point in their lives (…and purchase actual chemical-lab supplies for their coffee brews).
