FutureSnapshot triggers Rust Panic, PR Available
I am trying to get async working on WASM for Monty and I got a PR. I stepped through it but I dont know Rust, Typescript nor the NAPI. Solved it with TDD and found there were two issues. I have the plan, the commits (together its about 1.2k lines 700 of that is tests), etc. This does allow async to work on WASM through the tyepscript bindings. I have a compatibiltiy test suite that I use to run with native and WASM. And the async failures for WASM that failed are now working. I also added some stress tests. The code paths I was using work.
Q: Should I bother submitted the PR? If you guys know about this problem and your redesigning it; I will fork Monty and use that for a bit until you guys fix it. If your ok with me submitting a PR that I know nothing about I have validated works.
Below is LLM talk:
Bug 1 (prepare_current_task_after_resolve): Doesn't check frames.is_empty(). After a partial resume where load_ready_task_if_needed saves the task's context (draining frames), the next resume still claims the task is ready and pushes to an empty stack. vm.run() panics on "no active frame".
Bug 2 (resolve_future gather path): mem::take(&mut gather.task_ids) empties the gather's task_ids before checking completion. If the gather isn't complete yet, handle_task_completion later reads the empty vec, considers the gather vacuously complete, and panics on unfilled results.
Should I submit the PR and remove the excessive AI comments?
1 条评论