My application must draw one SVG chart, and the SVG support on Qt is quite bad although I can render my SVG there..
We do have a tray icon too, though.
So when you browse to our site via the desktop app, certain OS-only features turn on.
So it made a bunch of sense to use QtWebkit. I welcome the switch to QWebEngine though, if for nothing else in the hopes that the underlying browser code gets a update and the app will hopefully then feel more responsive.
Atm we use a mix of QtQuick/QML and some widget code mainly when wrapping QtWebKit it for some parts of the app furniture that need it.
Although the guts of our app requires native code (for clipboard monitoring and websocket message receiving), were considering ditching much of the UI in favour of a webapp.
The problem is that I know how far I can go with Qt Widgets, but on the same way, I would like to build something more "modern" and Qt Widgets may be quite limited..
My advice - just do it :-). If you want to build a cross-platform desktop app then go for QtQuick/QML/C++. QtCreator has a bunch of example projects. Use those as reference. The Qt docs are solid. Worst case - you will learn that Qt/C++.
I am not a big fan of PyQt/PySide, especially when starting fresh with Qt - you would have a hard time figuring out how everything fits and where the issues are.
[1] https://github.com/papyros/qml-material
Paraphrasing a previous comment:
I have used web-based task/todo lists (trello, asana, etc.) but I wanted something that could be used offline - my work does not allow storing company sensitive data on 3rd-party servers.
Started the project with Electron, using AngularJS and Angular Material. That allowed rapid prototyping with a slick UI and it was fairly easy to find additional Angular-based plugins. My main motivation in selecting Electron and Angular was to learn the two technologies (easier to learn something by doing a project in it). Screenshot: http://imgur.com/NZzEFKX
Had a few issues with the Electron app - big download size (50-70MB zipped), no print functionality, app felt non-native, very convoluted process to get it published on the Mac App Store.
Decided to change the tech stack to Qt/QML. I have published C++-based games, built on Cocos2dx. Wanted to try Qt to build a serious app.
Gumroad has been a slick experience - I have opted for the fully functional trial option and let the user buy a license to use beyond the trial period. Gumroad's onboarding (buy workflow) and tech (license key generation and validation) have been a revelation.
Note that the LGPL version of Qt doesn't contain everything the commercial version does. But there isn't a lot missing.
We had no choice technically-we need to use audio APIs on Win/Mac to get our job done.
Based on my experience, I would not recommend building a desktop app unless there is a technical reason to do so.
Data storage, building installers, signing your app for target OS's, and dealing with auto-update are all technical problems that I'd prefer to avoid if I could. Building cross-platform code is also a pain and, while totally doable, will slow down a team that is not familiar with it, I think.
If I really needed to build a desktop app in the future, I'd investigate building building a headless app that hosts a websocket, and use the user's browser to connect to a localhost websocket to communicate with it. (i.e., a technical choice freeing my from Qt)