Closed bmhowe23 closed 2 weeks ago
"Non-repeatable behavior is just a nuisance sometimes, but it can also apparently cause some of our Python notebooks to fail if all the dependencies aren't correct."
Basically we just don't need sorting but sorting based on dependencies of these tests, right?
Required prerequisites
Describe the bug
Our Publishing pipeline is not running tests in a repeatable order, which can cause non-repeatable behavior.
For example, one can compare the logs from the following runs
The
Testing photonics:
line appears later in the other run.Non-repeatable behavior is just a nuisance sometimes, but it can also apparently cause some of our Python notebooks to fail if all the dependencies aren't correct.
The root cause is likely all of the
find
commands invalidate_container.sh
. E.g. https://github.com/NVIDIA/cuda-quantum/blob/735e62b27a2f534278b225e53453ad131b1cbf0b/scripts/validate_container.sh#L130. Thefind
command returns results in a non-deterministic order (depending on how all the files were created), so its outputs likely need to be piped tosort
to achieve a repeatable behavior.Steps to reproduce the bug
See above.
Expected behavior
The publishing pipeline needs to run tests in repeatable order.
Is this a regression? If it is, put the last known working version (or commit) here.
Not a regression
Environment
Suggestions
No response