So, if not Electron apps, then what? | Hacker News Reader

So, if not Electron apps, then what?

27 pointsvijaybritto34 comments

I'm trying to get some data on building a cross platform desktop app. It's to help devs in my company by giving a GUI for few tasks. But I don't want to suck up their RAM. What are the other alternatives that come close to electron in terms of productivity while consuming low memory?

View on HN

Tk is pretty fast. It has bindings in multiple languages, though I'd suggest TCL. There is a starkit idea floating around, where people package a binary for each platform along with any other dependencies. It is much lighter than electron.

edit: to be clear, one clickable thing that works on mac, windows, linux, and... maybe arm? there is an androwish, but I haven't tried it.

Tbh I'm surprised React Native hasn't been brought to Desktop yet. It solves the problem with nice compromises imo, and gives a great native experience in the ui layer.

People are trying

  https://github.com/Microsoft/react-native-windows
  https://github.com/ptmt/react-native-macos
  https://github.com/status-im/react-native-desktop

But without top-level standardisation effort it will not work.

In my limited experiments with react native on the desktop, it was not any better than electron for memory usage

Qt, Gtk, Tk, wxWidgets, Delphi (whatever the open source one is). All have pros and cons depending on your needs but you should be able to throw something pretty serviceable together pretty quickly. The biggest catch for most is the reliance on c/c++, but if you want to be lean on memory then you probably want this anyway.

Some of these are heavyweight, others are just wrappers for the native libraries.

If this is a dev tool I wouldn't rule out a curses based TUI app either.

> curses based TUI app I really love these

FreePascal with Lazarus. You can create native apps (including mobile apps) all with one Object Pascal code base. If you want a commercial product and are on Windows you could always use Delphi as well.

Nothing really. If you don't want to sacrifice productivity then you go the electron way. Java/Swing has horrible UI/UX and doosen't offer any real benefits on RAM. QT is good if you are willing to pay for the licensing costs for commercial use or you open source your app. Everything else that is cross platform works, until it dosen't. Good luck debugging gtk on windows.

If you can spare the resources, create per platform apps. If not, then electron is your only headache less option.

> QT is good if you are willing to pay for the licensing costs for commercial use or you open source your app.

Qt is LGPL, you can use it commercially for free.

It sounds like this is for an internal app they don't plan to distribute, so even the GPL would be fine and not require them to open source it.

What about JavaFX instead of Swing?

JavaFX, especially when used with Kotlin and TornadoFX, is really nice.

We're now at the point where you can bundle your app, JVM pieces needed, into a single distributable for each platform. With Graal and whatever else on the horizon there might even be just plain compiled-to-native options coming too.

If you don't like the JVM ecosystem I think QT also has potential but it seems much more difficult and labor intensive to create a decent application.

The idea of using web views for the UI is valid IMO. The problem is that Electron brings a ton of metal with it (Chrome + Node).

A solution is to create your own Electron by creating the UI with HTML/CSS/JS and using the OS native APIs for the window and "back end" code.

Depending on your use case it might be feasible to do that yourself, or use a project like NodeKit.

https://nodekit.io/

What does the app do? How much RAM do the company machines have?

Personally, I'd take the hate about Electron here with a huge pinch of salt. RAM isn't expensive, developer machines tend to have a lot of RAM and you're only wanting to launch a single app. A minimal Electron app takes about 100MB of memory which isn't much if you've got e.g. 8GB of RAM.

I use TornadoFX daily and its a joy to use.

This looks great, thanks for bringing it to my attention!

How about this?

https://github.com/zserge/webview

C/C++/Golang webview. Build your UI in html/js and offload any heavy tasks to compiled languages. There's also a rust binding on github.

The sample app shows what you can do

Depending on what you're trying to do, Xojo might be a good choice: https://www.xojo.com/products/desktop.php

Others have mentioned Qt/QML, another solid choice.

Why not a web app?

Love2D with a GUI (lua): https://love2d.org/

miniSpere (js): http://www.spheredev.org/

What's the financial ROI effort spent to "save Ram" on developer's computers versus just adding hardware capacity to those same computers? The ROI should include productivity gained lost based on delivery schedule, the costs of ongoing app maintenance, the cost of finding developers in whatever alternative to electron is chosen, the value in terms of security and bug fixes that a mainstream project like Electron provides, and the economic value of having RAM on developer computers available for other business productive purposes. My gut assumption is that $200 of RAM upgrades across 50 developers is probably cheaper, but I could be mistaken. Good luck.

Not everyone can upgrade their RAM without switching to new hardware, and the company's probably not going to invest in that across their entire dev team just to accommodate one new internal app unless it's truly mission-critical. And heck, Mac laptops still max out at 16GB.

And it's not just RAM. Electron rendering is slowwwww.

I don't disagree that there may be a hardware upgrade. My point is that without running numbers increasing development, maintenance, and risk in order to "save RAM" is dogma not engineering. Rendering speed is just one consideration. That's why Slack.

That may be. But there's also no point in building something at all if your target audience's hardware won't run it or if it'll slow those machines to a crawl. There's a decent chance that the OP has no power over the pursestrings that would need to be opened for a hardware upgrade.

If it's for devs just stick a small server and thin HTML front end in a Docker container. Or just give them a CLI, I prefer CLI's over a GUI most times anyway.

If one is down with React, this project has promise:

https://proton-native.js.org

Looks interesting. Anyone tried this?

Seriously? Are your devs running MS-DOC with 640k of RAM? If not then don't worry about it.

A Java/Swing application would be easy to maintain long term, even decades from now.

Spending on the requirements, possibly Python and TKinter?

native apps.

I'm asking for productivity. I want to develop for 3 platforms and I don't have a lot of time!

I think that .NET Core have a stable UI kit.

Else, my favourite solutions are Kotlin/TornadoFX/Java/JavaFX.

> I think that .NET Core have a stable UI kit.

It does, Avalonia: https://github.com/AvaloniaUI/Avalonia It looks like a pretty promising project.

There was another one I believe I saw on HN a while ago but I can't seem to find it.

Avalonia is not stable and is in beta. I would not use it for production.

Reply on news.ycombinator.com