And the team is nice (I've met some of them at PyCon or FOSDEM).
The framework is pleasant to use, and there is a descriptive language, kv, which is really great.
Cross compiling may be painful though (I did it for Android) and the app loading time is a bit long, but it's working.
Some things may be missing in comparison to big frameworks such as Qt, there is no WebView for instance, and accessibility is unfortunately not as good.
It's overall a very good project and it's a pity that it's not more known and used.
I used Kivy once a few years ago for a device that had strict constraints on what it could run with the requirement to run the same code on desktop too (to display the same data). It worked very well for that.
It was not an elaborate app, so I cannot comment on how well it might work with something bigger, but it worked very well for what I needed.
It's pretty new though, and it's a bit rough around the edges still. The other issue is that almost every single package/library for pyside6 only supports QtWidgets, not QML. Meaning you wouldn't be able to use tons of libraries that make up the python qt ecosystem (for example, pyqtwidgets or vispy). Not a huge dealbreaker but it's something to keep in mind.
Yep I don't think there's a practical way to use qtwidgets on mobile, so it's not like Qt is treating pyside differently. It's just that the ecosystem doesn't really follow. I wonder if that's different on the c++ side. Is QML more prevalent in the c++ qt ecosystem?
Thanks for sharing your experience and links! Based on plyer's GH, it looks like you could use it to develop an iOS and Android app. If it supported more APIs (like Health Kit, accessibility), I would try it in a heartbeat.
Meaning Kivy is not yet a good choice for user-facing apps. It is so frustrating to see all the new UI frameworks and they fall apart if you just ask about accessibility features that should be absolute standard in 2024.
Is there a cross-platform accessibility library? I can tell you that developing a cross-platform GUI framework is a gigantic can of worms fraught with forced yakshaving, no documentarion, no support, and endless bugs across the stack even down into the OS and GPU driver stack. So in my opinion, everyone asking for accessibility features on every GUI framework announcement should get together and make a GLFW for accessibility libraries so that it can indeed be standard.
Thanks. I'll take a look. But it being in Rust makes me a bit hesitant. A lot of these projects in Rust seem to fizzle out as soon as people get bored with them once they become hard.
GUI frameworks in any language often fizzle out when they become hard. Its probably one of the larger cases of devs not knowing the massive undertaking they’re embarking on when they start.
Not quite what you're asking for, but Microsoft (my employer) has a free tool for checking web and Windows apps for accessibility best practices: https://accessibilityinsights.io/
The landing page is weird; it talks more about the funding for the framework than the framework itself. There's only one image showing UI, and the way its styled (cropped, tilted) makes me think its a stock photo, not a screenshot. The stock photo of a train right underneath isn't helping this perception for me.
And that's one of the main show stopper for me with kivy: it comes with very few built-in UI controls, so you have to code a lot of things yourself.
I much prefer Python to JS, but things like react native win because of the community libs you can install save you tons of time, and produce a better result.
This is especially true when you use a lot of tooling. I love jupyter, but installing it in a venv means pulling a lot of deps which will affect a lot what I can install.
Fortunately the Python community is much more serious about making deps that work together than the JS community, and the fact it works at all given the cartesian products of all the python modules is kind of a miracle and a testament to that.
Unfortunately, that's a problem that is unlikely to be solved in the next decade, so we all live with it.
The reverse problem is true for JS, and I see many projects shipping very heavy frontend code because despite all the tree shaking, they embed 5 times the same module with different versions in their bundle. That's one of the reasons for the bloated page epidemic.
I guess it's a trade-off for all scripting languages: choosing between bloat or compat problem. Rust and Go don't care as much, and on top of that they can import code from 10 years ago and it sill works.
However, and while I do know how hard it is to ship python code to the end user (at least if you don't use a web app), I don't think the version problem is the reason. We have zipapp and they work fine.
No the main reason iscompiled extensions are very useful and popular, which means packaging is solving more than packaging python, but a ton of compiled languages at one. Take scipy: they have c, pascal and assembly in there.
This can and will be improved though. In fact, thanks to wheels and indygreg/python-build-standalone, I think we will see a solution to this in the coming years.
My ideal situation is that the system should maintain authoritative versions of every package and version that is ever requested, and they should not need to be shipped. Multiple versions of a package should coexist.
When a package requests 2.1.1 it fetches it right out of there, installing from PyPI if it doesn't.
The same should be true of JS and even C++. When a C++ app's deb package wants libusb==1.0.1 it should NOT overwrite libusb-1.0.0 that is on the system, it should coexist with it and link to the correct one so that another app that wants libusb-1.0.0 should still be able to use it.
> Fortunately the Python community is much more serious about making deps that work together
This is very not true at least in ML. I have to create a new conda environment for almost every ML paper that comes out. There are so many papers and code repos I test every week that refuse to work with the latest PyTorch, and some that require torch<2.0 or some bull. Also, xformers, apex, pytorch3d, and a number of other popular packages require that the cuda version that is included with the "torch" Python package matches the cuda version in /usr/local/cuda AND that your "CC" and "CXX" variables point to gcc-11 (NOT gcc-12), or else the pip install will fail. It's a fucking mess. Why can't gcc-12 compile gcc-11 code without complaining? Why does a Python package not ship binaries of all C/C++ parts for all common architectures compiled on a build farm?
I'm assuming by system you mean OS, which is a terrible, terrible idea. Dev stack and system libs should not coexist, especially because system libs should be vetted by the OS vendor, but you can't ask them to do that for dev libs.
> I have to create a new conda environment for almost every ML paper that comes out
That's how it's supposed to work: one env per project.
As for the rest, it's more telling about the C/C++ community building the things bellow the python wrappers.
That causes 50 copies of the exact same version of a 1GB library to exist on my system that are all obtained from the same authority (PyPI). I have literally 50 copies of the entire set of CUDA libraries because every conda environment installs PyTorch and PyTorch includes its own CUDA.
I'm not asking the OS to maintain this, but rather the package manager ("npm" or "pip" or similar) should do so on a system-wide basis. "python" and "pip" should allow for 1 copy per officially-released version of each package to live on the system, and multiple officially-released version numbers to coexist in /usr/lib. If a dev version is being used or any version that deviates from what is on PyPI, then that should live within the project.
Actually conda creates hardlinks for the packages that it manages. Found this out a few weeks ago when I tried migrating my envs to another system with an identical hierarchy and ended up with a broken mess.
> but rather the package manager ("npm" or "pip" or similar) should do so on a system-wide basis.
I basically agree with this. With the caveat that programs should not use any system search paths and packages should be hardlinked into the project directory structure from a centralized cache. This also means that a dev version looks identical to a centralized version - both are just directories within the project.
Kind of, but not really. Nix is extremely complicated. Programs / projects including their dependencies is exceedingly simple.
Also, Windows is my primary dev environment. Any solution must work cross-platform and cross-distro. Telling everyone to use a specific distro is not a solution.
It is complicated... but honestly I have found claude 3.5 to just 'fix it'. So you hardly have to spend much time spelunking. You just give it all your dependencies and tell it what you want. It'll whip up a working flake in a few iterations. Kinda magic. So yeah when you can abstract out the complexity it moves the needle enough to make it worth it.
It’s not the researcher’s fault if the libraries they use make breaking changes after a month; proof-of-concept code published with a paper is supposed to be static, and there’s often no incentive for the researcher to maintain it after publication.
At this point, venvs are the best workaround, but we can still wish for something better. As someone commented further up, being able to “import pytorch==2.0” and have multiple library versions coexist would go a long way.
I install most tooling, including Jupyter, using pipx. The only thing I then need to install in the project venvs is ipykernel (which I add as a dev dep), and then create a kernel config that allows Jupyter to be run using that venv.
The problem I see a lot of JS developers having when they start using Python is they try to do the "import the entire world" strategy of development that's common in JS, and there isn't good tooling for that because Python just doesn't have that culture. And that's because it's a bad idea--it's not a better idea in JS, it's just more part of the culture.
Pick one package source. Stick with it. And don't import every 0.0.x package from that package source either.
There are obviously reasons to use more than one package source, but those reasons are far rarer than a lot of inexperienced devs think they are. A major version number difference in one package isn't a good reason to complicate your build system unless there are features you genuinely need (not "would be nice to have", need).
And it doesn't provide any way to use a link to any other package repository if you want to stick to vanilla pyproject.toml + build (the official build tool). So if you want to use the CUDA or rocm version of torch, for example, you have to add a direct link to the package. That means that you'd have to hardlink to a platform specific version of the package. There's no way to just make a package look at a non pypi repository to get the version you want otherwise.
So say you want to add pytorch, with GPU acceleration if it's possible on a platform. You want to make it multiplatform to some extent.
You can't add another index if you want to use vanilla build, as that's not allowed. You can add a direct link (that's allowed, just not an index) but that's going to be specific to a platform+python version. Pytorch doesn't even provide CUDA packages on pypi anymore (due to issues pypi), so you need to be able to use another index! You'd need to manually create requirements.txt for each platform, create a script that packages your app with the right requirement.txt, and then do it again whenever you update. Otherwise, I think the most recent advice I've seen was to just make... the user download the right version. Mhmmmm.
The other option is to use poetry or something like that, but I just want to use "python build . "...
But you can do that, obviously not with this syntax. It’s non standard but I have built programs that install all dependencies as a first step. It’s pretty trivial.
FWIW JavaFX is still great too BTW. I have a decent sized app that'll run fine with a 50mb heap, and you get native OS installers too. No web support though.
Does it appear native? And I keep hearing that even though swing is really old, it has better performance and that it’s used by IntelliJ. Not sure though
On Mac, the only thing that looks and feels native is the actual native Objective-C or Swift framework.
Everything else has subtle differences, and yes, your users will notice. You may as well just ship a janky Electron app like Teams. But if you want to impress your users with a native feel, it's the Apple toolkits or nothing.
One cross platform Python framework I found interesting is flet https://flet.dev/
It's powered by Flutter behind the scenes and familiar enough so that you can translate most things from Flutter/Dart tutorials to Flet.
I haven't used it and I'll most likely never will (Flutter developer trying to pivot to real native development), but it seems to have an active community, and in theory, it enables developers to write relatively nice looking apps with a very popular language.
Why do people keep doing things like investing in a framework, written by one set of developers, built on another framework, written by another set of developers, and acting like that is not insane? Unless both stacks are written by huge software companies that have an extreme vested interest in keeping them running, there is a huge risk of abandonment. Maybe Google just decides maintaining flutter is a pain in the ass and React Native is just good enough. Maybe the flet devs don't turn enough profit. Maybe flutter changes too radically and makes it difficult for flet to catch up. Such situations have happened again and again and again, and they will continue to into the future. What are we gaining here that is worth such an enormous risk?
> What are we gaining here that is worth such an enormous risk?
Well, Flet's landing page has a bunch of features and benefits. What is an "enormous risk" anyway? Depends on what you're doing. There are endless cases I can imagine where I just don't care at all about such a risk.
While the risks you mentioned are indeed valid risks, it doesn't automatically mean that these risks (however small or huge they be) outweigh the benefits the Flet project offers.
It's somewhat of a niche project, and certainly not for everyone, or for every project, but it's an active project with some real users.
Interesting. Looking for something like this to use instead of Flask, which is great on the server side but need something that’s easy to work with on the client side that provides a feature complete and nice looking set on controls for the client side.
Last time I looked at Flet, it gave the impression that you had to use the wrapped Flutter widgets, making it hard to use widgets and packages from pub.dev in a Flet application.
Kivy led to one of my first open source contributions in 2014. I submitted Kivy file support to Vim. It was done via patch diffs over email with Bram Moolenaar. It's still in Vim today[1].
Wrote a fairly complete POC in 2012 with kivy, it was able to render rather detailed floor plans, which was the most important feature of the POC since the idea was that, given the complexity, it should be written once and be able to run on multiple platforms with minimal changes, keeping in mind that mobile platforms were the priority.
Most impressively, it was running very well on a first generation iPad, not to mention Android tablets, and of course, Mac, Windows, and Linux workstations.
It was ultimately dismissed by the stakeholders because there was no way to render a web page inside the App, which was something kivy couldn't do back then.
Katrain uses Kivy and is problematic to package and upgrade because of it, on rolling release distros. When installing it with pip, it partially works.
I still use tkinter for apps that I make for my own use. Those are often for lab automation, where a primitive GUI lets me avoid needing 3 hands while running an experiment, and the support for live graphs is extremely useful.
For stuff that I'm likely to share with others, I've been moving over to Flet, because it lets me build things like webapps and stand-alone Windows apps. Both of those things can eliminate the hassles of installing Python apps as experienced by non-programmers.
I hate writing GUI code, and in both cases have written crude wrappers that let me throw together a functional dialog with tolerable defaults for everything. This approach works for "programmers who shouldn't trust themselves to write GUI's," similar to the old Visual Basic.
I think it's possible to use tkinter to develop even modern apps which don't look primitive, there are many open source projects which have done that including the IDLE editor itself, a part of core python project. The only issue is that it sometimes becomes excruciatingly difficult or painful to even perform some very little non-trivial things (like customizing a data-grid with a different style or having multiple tabbed pages) which are quite a simple task in other toolkits as they're already available in the form of a widget or component.
If ever I decide to leave tkinter, I think a viable alternative I will try is free-pascal[1]. It's a more comprehensive and well-thought toolkit for creating GUIs, more importantly it is also open source and cross-platform. It's the FOSS version of Delphi IDE.
For Kivy to stop being a niche UI framework with marginal adoption it must somehow tap into the central role of Python in the broader data science / machine learning universe.
Building data-centric cross-platform apps while staying (mostly [1]) within a single language ecosystem should be less friction and overhead than juggling multiple universes via API's, different runtimes and what not.
What kind of apps would benefit from this "single language" approach? For sure not the more open ended, exploratory data science type tasks. These are better delivered via notebook workflows, which, besides flexibility, enable better reproducibility and auditing. Also probably nothing that requires high performance interactive graphics.
But while not solving all UI problems for all people, there should be still plenty of relevant use cases where simplicity and fast prototyping give Kivy an edge when the task is to make algorithms and related tools available to non-technical users.
[1] ofcourse the actual number crunching might be done by yet another layer (typically C/C++) but that layer is essentially hidden from the data orchestation and UI integration that would be the Kivy app focus.
I have been following the progress of this project for the past two years, but unfortunately the community support does not seem to be ready to take this project anywhere. The most basic app store in-app purchase support is directly rejected. If an application framework cannot quickly help people get started, how can it get people's support?
I've kept an eye on Kivy for quite a while and prototyped a bunch of projects over the years with it, and I'm not a big fan.
It feels like it's at the wrong level of abstraction for... basically everything.
The Pong game demo is an example of this: if you're writing a Pong clone (or most video games) where you're going to be operating on a canvas, you don't need all the widget infrastructure that Kivy offers--you're better-served by something like PyGame.
On the other hand, if you want to build a UI using standard widgets, the widgets they provide out of the box aren't particularly fully-featured or even good--you end up doing a lot of hand-coding of functionality that could be included, and the defaults aren't particularly desirable, so you end up having to configure a lot of, for example, visual display settings.
As another user pointed out, their default widgets don't support accessibility meaningfully, and there are many other features, such as dark mode/color scheme support, which modern users expect and which you'll have to code yourself. Realistically, a lot of clients aren't going to give you funding for accessibility features, so the defaults are what most projects will end up with, and if it's me developing it, I'm doing accessibility on my own time, so I'd want this to be configuring what's largely already there, as opposed to what Kivy has: implementing it from scratch. In 2024 I'd view failing to support accessibility reasonably out of the box, as almost a moral failing, and certainly this is enough to discount Kivy from being used for any product intended to go to production.
There IS a fairly vibrant ecosystem of 3rd-party widgets (flowers; there's a "garden" metaphor in their branding for the ecosystem). But this comes with all the problems of a 3rd-party ecosystem: Kivy itself is probably large enough that it won't become abandonware in the forseeable future, but 3rd-party projects aren't, and there are large security and reliability risks to pulling in a bunch of small packages maintained by developers of various talent, intention, and funding. These are risks you generally have to accept for something unusual, but you shouldn't have to accept these risks for your bread-and-butter widgets like buttons and dropdowns.
If you're embarking on a project that benefits from using pre-built widgets like this, the framework I'd recommend is Flet. My experience with it has been overwhelmingly positive, and I've entirely switched away from PyQt for any new projects. The one criticism I'd give is that it doesn't really support multi-window, but that's something I'd avoid for most projects because multi-window support can never really be cross-platform, since mobile platforms don't really support windows as such.
All these types of frameworks, ones that have code to make their own UI are fine, but what I really want is to have a declarative UI, like HTML or XAML, bound with some user interaction event engine and access to the underlying OS.
Hopefully, we can have that someday, without the likes of Electron or whatnot.
I would like to write a react app, with all its progress in UI/UX and thousands of libraries, package them into a Python app and distribute it as a desktop application.
Kivy gives you a pretty good separation of logic and UI with its declarative yaml-like kv files. For me, it conceptually feels a lot like developing in React.
A) this is most European website I’ve ever seen. I couldn’t tell you exactly why… perhaps it’s the train subconsciously affecting me.
2) this is first time I’ve ever seen mobile included as part of “Cross-Platform”, that’s pretty awesome. We’re living in the future, friends! Tho it also makes me shudder at the thought of the phrase “QT app development”…
III) At this point, why not just use web? What is a “truly cross platform UI framework” other than HTML? I’m currently developing a site that uses TS in the frontend and Python in the back, and that seemed like a nice Unix-y division of labor. What am I missing?
I recently learned that Ireland isn’t really in Europe according to some Irish people, so I think this is final straw: I’m adding Japan to Europe. We may lack the technology to make it a physical reality (yet!) but their recent history and culture are quite European. I wish they were an option for emigration in a post-P2025 world, but AFAIK they are not at all interested in asylum seekers, American or otherwise ;(
Really? Where do they think it is? yes, its an island but no-one says Japan (or Sri Lanka, or Java) is not in Asia!
I have known a lot of Irish people (I mean either over here in the UK temporarily or first generation immigrants) and I have never heard anyone say that.
nice, I love casual exceptionalism in the morning.
> At this point, why not just use web
Because not everyone uses a "web stack". Moreover trying to learn a webstack is pretty hard. Sure JS is easy, but everything on top to make it more "productive" changes every 6 months. I mean yeah I could just use react, but Urgh, its nasty.
However the killer issue issue is hardware acceleration. Its possible to get kivy running at 60 FPS on a pi3. getting something to run at 60fps on web for the pi3 is bloody hard. Even though its python, it runs really fast, something not really possible in the browser.
I think a lot of people dislike html/web-based apps, and they are not as responsive in some cases as well. I have seen some mobile browser implementations that explicitly put large delays (hundreds of ms) into their touch handlers for example. You can see a similar delay in a side-by-side comparison video here https://www.youtube.com/watch?v=Z4CwVN9RRbE
Kivy's marketing seems to be targeting LOB apps. If I was going to develop one of those, I'd optimize for something standardized and easy to maintain (HTML/JS) vs. the performance benefits of a native UX or cross-platform framework.
Thanks for the reply! If touch handler delays are intentional then it seems like an advantage, not a weakness — who am I to disagree with UX experts employed by some of the top-paying tech companies or Mozilla? This is really a synecdoche of the overall situation, from my perspective: some really good devs want absolute control over every part of the stack, and to feel “close” to bare metal as they code, whereas the dummies like me appreciate that someone did a lot of the hard work for us.
In other words: how would that demo hold up if you asked them to recreate some popular react libraries, such as tables, graphs, and 3D simulations? When dealing with hard tasks (and network latency!) it seems like any platform-level delays would be quickly dwarfed by context-specific delays. And in the latter case, I’d rather have NPM than https://kivy.org/gallery.html !
PS did anyone else know NPM is now owned by Microsoft? We seriously need a revolution, or at the very least a figurative corporate guillotine. They own most of gaming, most of NLP/AI, most of dev tools, most of business/office computing, most of the OS market in general, most of… everything, it seems like. I’m just thrilled in hindsight that the windows phone failed!
For UI there is https://github.com/kivymd/KivyMD for Material design on top of Kivy.
And the team is nice (I've met some of them at PyCon or FOSDEM).
The framework is pleasant to use, and there is a descriptive language, kv, which is really great.
Cross compiling may be painful though (I did it for Android) and the app loading time is a bit long, but it's working.
Some things may be missing in comparison to big frameworks such as Qt, there is no WebView for instance, and accessibility is unfortunately not as good.
It's overall a very good project and it's a pity that it's not more known and used.