https://github.com/python/cpython/issues/121108 is tracking how Python currently can't build under WASI threads due to bumping up against memory limits. I tried to fix it and I keep getting it wrong (honestly, guessing the right memory limits for CPython has always been tricky).
I'm at the point I'm wondering if trying to keep WASI threads working with CPython is worth it? I'm willing to do it if people deem WASI threads worth supporting, but it will also mean me needing help figuring out what I'm doing wrong. But if WASI threads is kind of dead then I'm also fine ripping out the support in Python as it isn't being tested right now anyway.
If you want a quick fix then setting the maximum size of memory to 4G shouldn't be an issue since I think that works in both browsers and out-of-browser runtimes (although worth checking, I'm only sure about the wasmtime side of things).
Otherwise though my personal feeling would be that it's ok to drop threads for now. Threads-and-wasm isn't as ubiquitous as threads-and-native and many platform providers don't support wasi threads at all right now
wasi-threads is effectively a dead proposal, there are some implementations that work with preview 1, but the spec isnt compatible with components/preview 2. the shared-everything-threads core wasm proposal is the successor to that work, which is working on the underlying issues with threads in core wasm that need to be resolved in order to use them with components
(Why did this month old topic show up in my feed today? I guess I'm using zulip wrong)
Last updated: Jan 24 2025 at 00:11 UTC