There was a period where we thought the web browser could become a low level general purpose runtime and we'd install things like sqlite or ffmpeg as wasm bundles
I haven't dug deeply enough to say that dream is 100% dead, but I can tell you that ffmpeg.wasm is not a winning solution
The bundle size is bad enough but then runtime performance is a nonstarter. I haven't looked deeply enough to say if there are optimizations in ffmpeg's build or wasm runtime features to add which could improve the situation.
at least not any time soon, yeah.
which also gets to another problem with wasm: totally insular. there is no way to really know what's going on, because they don't talk about it
anyway i'm mostly just bitter because i bought into that early vision hard
and it didn't have to go this way. but it did.
and just to be clear, i do think server side wasm is also a good idea. but both were very cool conceptually.
Can you perhaps explain what you mean by "insular"?
From my perspective wasm is developed very openly, in public repos/meetings/talks. I'm not sure what more we can do, so suggestions are welcome!
(Apologies if I've misunderstood your point here. I read it a few times but am not sure I follow.)
i think y'all are right for the current moment, but on a longer time scale (5+ yrs, e.g.) i think we'll see wasm find more purchase. prolly not gonna be installing fmmpeg, tho, to paul's point)
now with sqlite's official opfs (origin private filesystem) based wasm implementation, it's a lot faster, only being about 2-4x slower than native for large amounts of queries sqlite-wasm-opfs.glitch.me/speedtest.html
