Problem
generate_uuid(...) currently derives Perfetto IDs with Python's built-in hash(...):
return hash(f"{workflow_run_id}_{attempt_number}_{job_id}_{step_id}") % (2**64)Python hash randomization means the same workflow run can produce different IDs across separate Python processes. That makes generated traces harder to compare, cache, or diff, and can make tests flaky if they ever assert specific IDs.
Suggested fix
Use a deterministic hash such as hashlib.blake2b(..., digest_size=8), sha256(...).digest()[:8], or UUIDv5-derived bytes, then convert to an unsigned 64-bit integer.
Relevant file
src/github_actions_tracing/main.py
Problem
generate_uuid(...)currently derives Perfetto IDs with Python's built-inhash(...):Python hash randomization means the same workflow run can produce different IDs across separate Python processes. That makes generated traces harder to compare, cache, or diff, and can make tests flaky if they ever assert specific IDs.
Suggested fix
Use a deterministic hash such as
hashlib.blake2b(..., digest_size=8),sha256(...).digest()[:8], or UUIDv5-derived bytes, then convert to an unsigned 64-bit integer.Relevant file
src/github_actions_tracing/main.py