ligurio / molly / 34514072576

92%

Build Type

push

github

Committed by ligurio

Commit Message

molly: propagate worker errors from a thread pool to run_test

A thread pool silently swallowed worker failures: pool:start() always
returned true, the coroutine scheduler ignored the result of resume(),
and errors thrown in fibers were never propagated. As a result
run_test() reported success even when open(), setup(), teardown() or
close() failed, or when a worker crashed.

The thread implementations now report a worker outcome:

- thread_fiber.join() returns the fiber result; a fiber error or a
  worker's (false, err) return is converted into an error;
- thread_coroutine's scheduler records the result of every resume()
  and join() returns the recorded outcome.

The thread pool aggregates the first error across threads: join()
returns (ok, err) and start() returns nil, err instead of raising.
runner.run_test() already propagated such failures, so DB setup and
teardown problems and worker crashes now fail the test instead of
being silently ignored.

Coverage Stats

25 of 29 new or added lines in 3 files covered. (86.21%)

11 existing lines in 4 files now uncovered.

713 of 779 relevant lines covered (91.53%)

135.45 hits per line