Tauri binding for Python through Pyo3 | Hacker News Reader

Tauri binding for Python through Pyo3

github.com

168 points0x199756 comments

Nice. It might be worth copying some of the introductory text from the Tauri package explaining what this does. Otherwise a person to lands on the readme gets a lot of technical detail about how it is built without any idea what it actually does.

The first word in the README (other than the title) is a hyperlink to Tauri.

Going further, what can I build using it? Basically, can I use Python on a Tauri project or can I use Tauri on a Python project?

Tauri in Python project

Python, along with C++ enjoys the benefits of Qt. You can build great stuff with cross-platform Qt. And it won't be webview(Electron / Tauri) based.

What would be the benefits of this? I can't think of any yet.

This benefits people and teams who prefer to build UIs with html/js/CSS.

Avoiding Qt licensing

Quite a few options there. Never really a problem.

I worked on a team the switched v2 of our product from Python + QT to Tauri specifically to get away from QT. If you have any kind of specific design requirements, QT is pain. The ability to use the frontend web stack is a huge win especially for a team that already has frontend developers.

Rust was a nice bonus.

Doesn't qt also offers a webview solution?

It’s very basic. Good enough for oauth flow but not much beyond that, definitely not to host your entire UI. Also I don’t see what the point of that would be.

My point it's that you said you changed from QT to Tauri because of being able to use frontend frameworks but QT also allows you to do the same. Anki does something like that, but since you mentioned rust that might be the reason of the change to Tauri.

But why would you pull in an entire UI framework, QT, just to use its webview? If it’s not clear, these are local apps we are talking about. The webview loads local content. This is what tauri does.

Did you evaluate QML/Quick?

https://doc.qt.io/qt-6/qtquickcontrols-index.html

I've been building a Python+Qt+PyInstaller app for a decade (ActivityWatch) which we are migrating to Tauri for a wide range of reasons. Perhaps the largest one is easy cross-platform packaging/builds (which is a lot of work with PyInstaller)

Web rendering has a rich ecosystem of tooling around a vast level of support for styling, layout, localization and accessibility support. By extension, it's one of the largest baselines in terms of knowledge, support and talent with at least some experience with.

This, of course, comes at a cost of performance as CSS alone is very complicated and has a lot there to support. On the flip side, there are many things that are easy to do with HTML+CSS that are much more difficult with native UI frameworks. Just the reflow support for multiple sizes/scaling in browsers is hard to match.

The benefit is that you get to use a web-framework to write your UI... The trade-off is that the world gets another app that probably could have been a website.

Since Tauri is just a thin wrapper over the system webview, what's the point of having a wrapper over a wrapper?

I don't think the Python ecosystem was lacking in browser wrappers up till now.

"a thin wrapper over the system webview"

That is very complicated if you take into account also Linux, Windows, iOS, MacOS, Android support, and related utilities like tray icon, etc. There are other efforts, too, but they are also wrappers. Like this C/C++ implementation https://github.com/webview/webview that targets only desktop operating systems.

I don't think the point is just about Python - this means you can use JS front-end with a Python backend for a local app.

Tauri also has their own experimental Servo engine: https://v2.tauri.app/blog/tauri-verso-integration/ So it's not necessarily a wrapper over system webview.

Verso just got archived :(

    As the project progressed, multiple significant revisions to Servo were released, and the Verso browser was unable to keep pace with these updates due to limited manpower and funding. Therefore, we will be archiving the repository for now and look forward to a future opportunity to revitalize the project and continue contributing to the Servo ecosystem. [0]

[0] https://github.com/versotile-org/verso

> Since Tauri is just a thin wrapper over the system webview

This isn't true. It's a thing wrapper over various things, including that.

Even if it were true - so what? There are probably 30 nested wrappers you could name even if you started at WebView as the outermost wrapper.

Not to distract from Python or Tauri, but lately I’ve been using Wails[0] which is similar to Tauri but with Go instead of Rust.

There are pros/cons primarily being Tauri seems to allow creation of mobile apps which Wails doesn’t, but overall it’s been fantastic since I find Go to be an easier language for me.

[0] https://wails.io/

Yep I've gotten basic Tauri stuff to run on Android.

It runs fine but, speaking as someone with no prior Android experience, once you need to leave the "Everything is in the view" model, it gets confusing. But that's more Android's fault than the framework's

I was exploring the idea of using it for personal projects, mostly monitoring of sensors with data coming from network connections (some in raw binary format, not text), control of remote relays and actuators, etc, thus the need for self updating graphic primitives for gauges, indicators, leds, switches, displays, etc. Does Tauri allow that? Asking because although I have had a developer experience on other traditional languages/platforms (C, Delphi, lately some Python) I have none in coding for mobile platforms, web development and Tauri itself.

Interesting how much cross-pollination is happening in the Python ecosystem with Rust.

I think the NiceGUI example is good but quite advanced, might be beneficial to contact the teams from Reflex or FastHTML, because if you could use PyTauri to create potential local apps with those popular frameworks, it could be a big win for them and that can help with marketing around the project.

Rust and Python have very compatible ideas on a lot of topics. For example both think that a developer writing normal code should not have to worry about null pointers or be able to cause a use-after-free. It's just that Python achieves that with runtime costs and Rust with compile-time costs and a complex type system. So developers of one liking the other makes a lot of sense to me. And pyo3 is an extremely convinient way to call between the two languages

> For example both think that a developer writing normal code should not have to worry about null pointers or be able to cause a use-after-free

Like 99% of all languages currently in use.

These things have long been solved way before Rust even existed. Rust has only filled the small niche of cases where you can't or don't want to use automatic garbage collection.

> complex type system

Python's type system is orders of magnitude more complex. Dynamic type systems are crazy powerful, this is why Typescript is such an complex beast.

Rust has static type checking, which is what you mean. Which also means that Rust is limited to the types that are can be expressed and are decidable with that system, while Python allows you to do... whatever and types are only checked at runtime or with external tooling.

Python's type system is easier to use but more complex. Rust's is simpler but harder to use. I know people forget the difference between complex and hard but it is an important one.

A better reason Python is Rust are seen together is that Python is an excellent glue language. Same reason people like C and Python. Plus both Rust and Python are pretty trendy these days.

> Like 99% of all languages currently in use.

Yes, but 90% of languages don't make a good complement to python. If you want to have a language to use in combination with python you want something with fast c-interop, which most garbage collected languages can't offer

> Rust has static type checking, which is what you mean

Maybe I shouldn't have used the word complex. But I did primarily mean that Rust uses the type system to ensure "safety", while Python primarily uses other methods. Yes, Rust doing static type checking is also an important component of it.

> A better reason Python is Rust are seen together is that Python is an excellent glue language

That would qualify any language as fitting well to Python. Which I don't agree with. C fits well to Python because it is a good complement: Python is slow to execute and easy to write, C is fast to execute and hard to write, and communication between the two is fast. Rust fits even better to Python because it's easier to write than C (well, easier to write correctly) and because interop is fast and convinient

> Like 99% of all languages currently in use.

That's true about memory, but nearly all mainstream programming languages have problems with null pointers: Java, C#, Go, etc.

This is mostly solved in C# - https://learn.microsoft.com/en-us/dotnet/csharp/nullable-ref...

So why can't Python call system WebView directly? Is passing through layers of Rust really needed here?

Abstractions are rarely "necessary" in a strict sense but since Tauri already solves the problem of doing cross-platform webviews, why not reuse that work?

I've used pywebview in the past ~2017. Works well.

https://github.com/r0x0r/pywebview

I don't understand why people like Tauri - the fact it uses the system web browser completely destroys the main advantage of Electron: that you can test it locally and be absolutely sure that it will render like that on any other system since the browser is shipped with it.

I agree. I don't mind VS Code or some app a person did eats 300 MB of RAM and is Electron if it does the job done. By the way good luck implement something like rtl, i18n, text select and right click context menus in you favorite C/Go/Rust/ImGui/ImmediateModeWhatever library.

Wanna switch between Arabic, Chinese, English in a textarea or input or the whole app? Trivial in Electron. Again, good luck with that in any other environment.

Electron is superior for any text/form apps. HTML/CSS/JS are truly magical if you dive deeper and for any form-like classical type of crud apps there is really no better option.

With our computers getting more RAM and disk space every few years - especially compared to AI needs, Electron is actually super lean compared to those AI llms models. Funny enough, LM Studio is an Electron app ;)

> With our computers getting more RAM and disk space every few years - especially compared to AI needs, Electron is actually super lean compared to those AI llms models. Funny enough, LM Studio is an Electron app ;)

This is a phenomenally bad take. This is exactly the thought process that has led to the insane software bloat problems we're dealing with now (with Electron as one of the worst offenders).

> good luck implement something like rtl, i18n, text select and right click context menus in you favorite C/Go/Rust/ImGui/ImmediateModeWhatever library.

> Electron is superior for any text/form apps. HTML/CSS/JS are truly magical if you dive deeper and for any form-like classical type of crud apps there is really no better option.

I don't think you understand what Tauri is.

For a lot of people the main advantage of electron was just being able to use the webdev stack for a desktop app. Tauri makes it less portable but is less bloated. Different tradeofs I guess.

Also: I think it’s kinda funny that Tauri is basically a very straightforward of example of trading developer comfort for benefit of the user, and you can’t imagine people using it.

Indeed. It's sacrifices engineering cost for customer experience.

I have never seen a Tauri app that was significantly less bloated than Electron. Can you share any?

The bundle size is, by definition, much smaller as it doesn't include the browser engine.

It's not for the benefit of the user when you get dozens of issues in Github because webkit-gtk or WebView2 changed something.

If you stick to good old HTML5 with CSS3, without cutting edge Blink-only features, you should be safe.

> the main advantage of Electron: that you can test it locally and be absolutely sure that it will render like that on any other system

That's not the main advantage of Electron. The main advantage of Electron is being able to use web developers to build your desktop software cross-platform for much cheaper.

because most Electron apps are not that complex. Just few forms with some buttons.

Do we have an open source Playwright for multiple mobile browsers?

I have drafted a photo gallery app in Rust / Tauri, using a JavaScript framework in the frontend. The backend can read directories and files directly, and because the backend and frontend are in a single process, the backend simply passes a file handle (path string possibly) to the frontend. In contrast Electron has to send the image file between processes. I started with Electron and I think that was the point I shifted to Rust / Tauri; seeing the images display immediately was a revelation. Rust / Tauri has the advantages of a desktop app, and I have the option to use the frontend as a web app also.

This Python binding (pytauri) is interesting too - I have colleagues with Python functionality they want to surface on the web, and this would give the possibility of running as a desktop app also - good for large datasets.

There's an entire separate copy of Chromium attached to every single goddam Electron app I install. It's completely insane. It's not at all worth the consistency you point out. Far too high a price.

The Tauri team is doing God's work. Electron was a fine enough idea, but I can't wait to see it improve or die. Imagine Electron supporting cross-compilation for mobile OSes. They won't close that particular gap with Tauri any time soon.

The times of browsers having weirdly different rendering behavior are mostly gone, in my experience. I'm sure ~98% of Electron apps that expect Chromium would render just fine/same under WebKit as well.

You mean the laziness of developers to write standard Web code and rather help Google in their plans to take over the Web?

Tauri also provides an experimental Servo web engine to follow Electron: https://v2.tauri.app/blog/tauri-verso-integration/

What is Tauri?

Electron.

Electron* but without the bloat of Chromium

A rust-based alternative to electron that leverages the system’s web view instead of a full copy of chromium.

Reply on news.ycombinator.com