Deep usage feedback on flet

[Qye-Leisure]

I am a novice programmer who only knows Python, and I have no knowledge of front-end technologies like HTML, CSS, etc. When choosing a front-end framework, I compared several options: PySide (the documentation was hard to understand), Streamlit (limited customization and poor performance), and Reflex (an excellent framework, but deep customization of components still requires front-end skills). Eventually, I came across Flet—pure Python, highly customizable, cross-platform across multiple devices, and high-performance (as advertised at the time). It immediately won me over, and the documentation was clear (version 0.28). So, I decided to use Flet as my primary framework and have been using it for about 3-4 months. The project I am developing is similar to Coze or n8n, a workflow automation tool, and also requires some video editing functionality. The biggest issue I’ve encountered with Flet is performance. However, by reducing refresh frequency and scope, and directly refreshing the page, most performance issues can be resolved. As long as it’s not about native or frame-level refresh performance, it’s generally sufficient. In terms of custom components, using Canvas and GestureDetector in Flet allows for the customization of most components, but some special components may be challenging. For example, creating an in-place rich text editor (like a code editor or advanced notepad) or a real-time video previewer (for video editing software) might not work well. For the former, I’ve tried many methods but still can’t capture input method content through keyboard listeners. For the latter, I haven’t tried it yet, but even using Canvas for real-time frame rendering might not work due to the high refresh frequency. I believe that in GUI development, if these two issues can be resolved, Flet would be an almost complete and versatile framework, capable of handling most application scenarios. I really like Flet and don’t want to learn another framework.

Summary: I really like Flet and hope it can continue to improve. In my opinion, Flet is already sufficient for many scenarios, but rich text editing and real-time video rendering preview are its current shortcomings. If these two issues can be resolved, Flet could truly become a mainstream framework. Of course, for the software I’m developing, some features may need to be compromised because the current version of Flet cannot meet these requirements, which deeply saddens me. I hope Flet can become even better (I really don’t want to learn anything new—it gives me a headache; save this struggling learner, please). My suggestions are:

1. The documentation should revert to the old style, organized by functionality or needs, rather than the new alphabetical sorting, which isn’t as user-friendly. Unless you’re already very familiar with all Flet components, the alphabetical order might work, or both layouts could be retained.
2. The documentation should include notes on performance optimization, such as reducing refresh frequency, refreshing all necessary elements at once, and avoiding direct page refreshes. This would help developers avoid pitfalls and improve the development experience.
3. I believe Flet’s core should be similar to Flutter—write once, run anywhere. So, Flet’s main goal should be to quickly achieve multi-device and multi-scenario support while maintaining simple, easy-to-understand syntax and high customizability. First, ensure Flet can be applied to most scenarios, then focus on performance, and finally, improve development experiences like visual development tools.