To me, issues such as large memcpys/system calls on context switches, having tasks not be able to compute concurrently with random other tasks, possibly having to autobox everything, etc. sound like they would spawn equal or greater complexity and performance loss than just using split stacks.
If you really wanted to avoid wasting memory, couldn't you allow new tasks to steal memory below the stack pointer of swapped-out tasks, changing that task's stack bottom pointer in the process? _______________________________________________ Rust-dev mailing list Rust-dev@mozilla.org https://mail.mozilla.org/listinfo/rust-dev