-
-
Couldn't load subscription status.
- Fork 40
Description
Both @Chemaclass and I found it odd that if we left PARALLEL_RUN empty in the .env file, it would run in parallel. This is because in the code it is evaluated as PARALLEL_RUN = true to run in parallel. However, the env_configuration.sh defines PARALLEL as true if it's empty.
Both @Chemaclass and I found it more intuitive that if it's empty, it should run synchronously (in the vast majority of languages, the absence of value/null/undefined is evaluated as false).
If we wanted to change this behavior, I believe we should invert the conditions in the code (i.e., use PARALLEL_RUN != false instead of PARALLEL_RUN = true) so that the code expresses a more realistic behavior.
For now, we've changed it so that the GitHub actions work correctly due to issue #39.
Another distinct matter is whether we want the GitHub actions to run in parallel (once the issue #39 is resolved). In that case, I think it would be interesting to have a separate .env for GitHub actions from the .env.example.
To summarize, I want to discuss three points in this issue.
- What behavior do we expect when PARALLEL_RUN is empty?
- My opinion is that it should be evaluated as false.
- If we want empty PARALLEL_RUN evaluates as true, do we invert the code checks so they don't lead to confusion when reading them?
- My opinion is that they should be inverted if we decide that an empty PARALLEL_RUN evaluates as true.
- How do we want it to launch in GitHub actions and how do we proceed to do it?
- My opinion here is that they should run in parallel as soon issue Total assertions count not working when parallel run enabled #39 is fixed, and we should have a separate .env from the .env.example for GitHub actions (i.e. .env.github).
What is your opinion?