Daniel Danciu

I can add a test that builds 2 graphs in parallel (using phases), but I won't kill the process as that causes the test to be undeterministic. On Tue, 8...

Ok, I added a test that builds 2 graphs at the same time. First it builds phase 1 for both graphs, then it builds phase2 for both and asserts that...

Flakey/nondeterministic tests are a pest. There is general agreement about this in the testing community, just google "flakey tests". Your argument that the expected result is well determined doesn't mean...

Flakey is a test that: - passes most of the time, but in relatively rare conditions it fails - the conditions under which it fails are unknown and cannot be...

This discussion is getting surreal. That was just a thought experiment. To the best of my knowledge, any build that finishes with exit code 0 builds a valid graph. As...

In my case, making sure that `nvcc --version` and `nvidia-smi` were both at the same version (11.4 in my case) fixed the problem. When `nvcc --version` was at 11.7: ```...

We are also seeing this issue with a docker build that depends on multiple packages from the same git repository. I suspect that as more and more people adopt the...

Note that in addition to deleting the data in the database, the inputs/outputs associated with the execution (stored in S3/GCS) would also have to be removed. I am a bit...

This issue is still very much not resolved, despite having been marked as resolved. Our logs are absolutely illegible, because this message appears millions and millions of times: Note that...

On a different note, a better way to filter out these spurious error messages without affecting actual legitimate error logs is to do something like: ```python class _FilterCallback(logging.Filterer): @override def...