Threads and the GIL

Opening your lesson

Two workers can each read the same counter and then overwrite one another's increment. A shared interpreter lock doesn't turn that multi-step business operation into an atomic transaction. You need an explicit invariant and a synchronization plan.

These exercises replay supplied interleavings and lock dependencies. They don't start operating-system threads, which the browser runtime doesn't provide. The workload rule distinguishes ordinary GIL-enabled CPython from free-threaded builds, and distinguishes CPU computation from waiting for I/O. Keep those assumptions in the answer. A correct trace model explains a possible failure without claiming it measured a real scheduler. You also check the circular lock dependencies that can leave workers waiting forever.

import json

def process_payload(payload: str) -> str:
    data = json.loads(payload)
    result = []
    for job in data:
        if job['kind'] == 'cooperative-io':
            result.append('async')
        elif job['kind'] == 'blocking-io' or job['free_threaded']:
            result.append('thread')
        else:
            result.append('process')
    return json.dumps(result,separators=(',',':'))

Distinguish cooperative I/O, blocking I/O, and the build-specific CPU constraint. First distinguish cooperative I/O from blocking I/O, then consider CPU work and the interpreter build. This is a checked planning rule, not a measured claim about scheduler speed.

Choose your character

Your Work Gym account