Closed mikelsr closed 1 year ago
The ideal thing would be to replace the implementation with an anchor now, but the current one provides performance benefits already.
Quick feedback: the current approach looks reasonable. I would just caution against over-engineering this feature for the time being, because we're eventually going to want to integrate this into bitswap. I would expect significant parts of this to get rewritten.
But again: looks good. I'll finish reading and then hit the "merge" button if I don't see any isssues.
Depends on #146.
This way we can avoid the overhead of sending the entire bytecode every time we want to spawn a new process, as long as that process has already been spawned at least once on the executor.
The ideal thing would be to replace the implementation with an anchor now, but the current one provides performance benefits already.
I'll work on a better version later on.