Can already see a nice web based AI app builder on the way, also can be used to quickly try apps before pushing to the device as interactive app store.
Please watch out some qemu targets for Cortex m0 or m3, increasing ram from linker out of supported ranges will cause random crashes. Would love to contribute if you are looking for people to crash some issues.
More that browsers have gotten sprouted downwards and obtained all sorts of low level access than older browsers didn't use to have. I believe this is partially why tools that were meant to just render markup have gotten so complex to build that a small team of devs is not enough to build a modern browser anymore. And by that I mean from scratch, not just piggybacking on Chromium or Gecko.
> I believe this is partially why tools that were meant to just render markup have gotten so complex to build that a small team of devs is not enough to build a modern browser anymore.
Isn't it basically the opposite? The hard parts of the browser are layout, styling, and multimedia stuff that goes into rendering markup compliantly. Then there's the infinite sink of optimization work for a JS engine, the high-level scripting language for that markup. The low level access that something like this emulator use is comparatively easy; a WASM runtime and Canvas blitting pixels from some shared buffer.
Or am I mistaken that a WASM engine is much easier to build than a performant JS engine?
Can already see a nice web based AI app builder on the way, also can be used to quickly try apps before pushing to the device as interactive app store.
Please watch out some qemu targets for Cortex m0 or m3, increasing ram from linker out of supported ranges will cause random crashes. Would love to contribute if you are looking for people to crash some issues.
Awesome effort btw.
I'm blow away at what Devs are able to do within a browser nowadays.
You could have done the same with plugins in the past.
And now we don't need to use them and expose ourselves to security vulnerabilities. Win win.
More that browsers have gotten sprouted downwards and obtained all sorts of low level access than older browsers didn't use to have. I believe this is partially why tools that were meant to just render markup have gotten so complex to build that a small team of devs is not enough to build a modern browser anymore. And by that I mean from scratch, not just piggybacking on Chromium or Gecko.
> I believe this is partially why tools that were meant to just render markup have gotten so complex to build that a small team of devs is not enough to build a modern browser anymore.
Isn't it basically the opposite? The hard parts of the browser are layout, styling, and multimedia stuff that goes into rendering markup compliantly. Then there's the infinite sink of optimization work for a JS engine, the high-level scripting language for that markup. The low level access that something like this emulator use is comparatively easy; a WASM runtime and Canvas blitting pixels from some shared buffer.
Or am I mistaken that a WASM engine is much easier to build than a performant JS engine?
> Or am I mistaken that a WASM engine is much easier to build than a performant JS engine?
Absolutely. It's very limited, and designed for purpose, rather than somewhat by mistake (like JS): https://github.com/sunfishcode/wasm-reference-manual/blob/ma...
08:46:37.125 [err] [fps] 4.0 (12 frames in 3.0s) 08:46:40.225 [err] [fps] 1.0 (3 frames in 3.1s) 08:46:41.224 FIRM | * ASSERTION FAILED: ASSERTN 08:46:41.443 [err] DEBUG post-reset: halted=0 stopped=1 R13=0x20002170 R15=0x080001b4 thumb=1 08:46:42.123 [err] clktree_recalc_output_freq: Clock PCLK1 output frequency (32000000 Hz) exceeds max frequency (30000000 Hz).
found an assertion just by clicking up/down. doesn't seem too stable.
Yeah that happens when trying to go up into the timeline.
Any way to load a local watch-face? Am hoping there is and I didn't see it. This would be very helpful for testing.
Should I be getting more than .2 FPS?
I'm not even getting that - its stuck at: 10:26:18.027 Downloaded SPI flash: 16777216 bytes 10:26:18.027 [config] icount: off 10:26:18.028 [status] Loading QEMU WASM module (17MB)...
Cool cool cool
Pretty impressive!
Probably could be made faster by:
- using native exception handling
- getting rid of asyncify (but it would require JSPI)
See my experiments with TempleOS here: https://zb3.me/qemu-wasm-test/