4 ms·
> My initial approach was to compile the client into WASM. This seemed promising as it would let me leverage my expertise in Go development. However, I encounte
by dxdm 3y ago
> My initial approach was to compile the client into WASM. This seemed promising as it would let me leverage my expertise in Go development. However, I encountered several limitations that would have necessitated substantial modifications.
I'd be interested to read more about this.
- owulveryck 3y agoHello, In a glimpse: The main issue was with the gRPC library. The support is very limited by now. Then the JPEG compression is slow in Go and it is CPU intensive. And finally, I even if I could generate the Mjpeg stream, how would I display it? Then I though about the “canvas” mechanism, but I could not address the backend of the canvas without heavy copying between wasm and JS. And remember the size was 2.5Mb. Anyway, I though that relying on wasm would make me implement a lot of image primitives that are natively accessible in JS (for example image rotation). Hope that answers your question.
- dxdm 3y agoI haven't really looked into wasm yet, and am wondering what constraints one might run up against for interesting practical applications like yours. You satisfied my curiosity at the perfect level of detail, thank you very much for taking the time!