Fixes #1700

• 1700

When instantiating steps, creating a custom ocelot config file name. Finally deleting the newly created config file when disposing the Step instance.

Proposed Changes

• Since, by design, a new instance of Step is instantiated for each fact/theory, we need to ensure that the corresponding fact/theory only uses its own ocelot configuration file, without interfering with the configuration files of other facts and theories.

Wow! Cool! Let's make 10 PR builds of 10 commits being added to this PR, to get some statistics on stability of this fix.

Just commit a small change of Ocelot core, not tests, and the 2nd commit should be revert commit of previous one. So, there will be 5 commits and 5 revert-commits.

Good plan?

@raman-m no way, that's really funny ;-) 6 out of 10... I can't reproduce that on my machine

@raman-m Ok, a bit better now, 9 out of 11...

@raman-m why did you remove xunit.runner.json?

I've reverted commit 9fac778 Sorry!

It seems your solution doesn't work for some reason... That's strange... It should...

Have a rest today! 😉 We will continue tomorrow, this week...

@ggnaegi commented

why did you remove xunit.runner.json?

Because you've disabled parallel execution for all tests by default.

@ggnaegi commented

why did you remove xunit.runner.json?

Because you've disabled parallel execution for all tests by default.

Yeah, I tried...

@raman-m 11 out of 12. It's not that bad...

@raman-m we might have a solution...

It seems this fix works! 👍 What was the root cause of failing? I don't think it was random port finder... 🤣

Do you confirm that we can still use parallelization of all tests?

Do you confirm that we can still use parallelization of all tests?

@raman-m Yes, we still have an issue with Open_circuit_should_not_effect_different_route though... That's the test failing at the minute. I can reproduce that on my computer, it fails after 20 minutes.

Open_circuit_should_not_effect_different_route

Forget about this test now. It is out of the scope of linked bug-issue. 🆗 I'm going to review...

Open_circuit_should_not_effect_different_route

Forget about this test now. It is out of the scope of linked bug-issue. 🆗 I'm going to revi

It seems this fix works! 👍 What was the root cause of failing? I don't think it was random port finder... 🤣

@raman-m No it wasn't, but the random port finder code doesn't seem to be thread safe, since the concurrentbag would allow identical values to be stored and the method can be accessed by several threads concurrently

@raman-m imho ok to merge

Revert "commit #h5" 6e4b8a4

@ggnaegi #1706 FYI

@ggnaegi Congrats! We did that! Thank you very much for your efforts during this Sunday's evening!