Chain previous exception in AssertionFailedError

[mondrake]

Fixes #5068

[sebastianbergmann]

Can you please add a test that fails without and passes with these changes? This probably needs to be an end-to-end PHPT test.

[mondrake]

Will try. Do we have to consider traversing the exception chain up to the root one to get the trace at each re-throw, or just getting trace of the previous one is enough?

[mondrake]

I cannot find a way to determine from within the exception whether the originating test was being run in isolation or not, so I added a parameter to the exception's constructor to indicate that context. Probably not a very suitable solution, some help would do.

[codecov]

Codecov Report

Merging #5069 (81cfbde) into 9.5 (4f2a753) will decrease coverage by 0.02%.
The diff coverage is 64.70%.

@@             Coverage Diff              @@
##                9.5    #5069      +/-   ##
============================================
- Coverage     83.78%   83.75%   -0.03%     
- Complexity     4595     4602       +7     
============================================
  Files           273      273              
  Lines         11430    11447      +17     
============================================
+ Hits           9577     9588      +11     
- Misses         1853     1859       +6     

Impacted Files	Coverage Δ
src/Framework/Exception/AssertionFailedError.php	68.42% <64.70%> (-31.58%)	⬇️

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[sebastianbergmann]

Please fix the issues identified by PHP-CS-Fixer and Psalm.

And may I also kindly ask you to properly configure your real name for Git's user.name setting?

[mondrake]

I fixed the code style issues.

I will not disclose my real name in a public environment, sorry. I have committed with name + surname's initial, which is as far as I will go. Thanks for understanding.

[sebastianbergmann]

Thank you for your contribution. I appreciate the time you invested in preparing this pull request. However, I have decided not to merge it as I do not believe this solution is the right one. Unfortunately, I currently do not have a better one.

I have created a branch that has a cleaned up version of your test case. This should serve as the foundation for future work on this issue.

In the meantime, I strongly recommend to change your implementation of onNotSuccessfulTest() to not rely on exception chaining. AFACIS, all that needs to be done is mapping the information from Mink's ExpectationException to PHPUnit's AssertionFailedError.

[mondrake]

Agree this was not viable. Biggest problem was impossibility to determine if a test was run in a process isolation context from within the exception.

Looking forward to a more sustainable solution. Let me know if I can help.

Thanks

[sebastianbergmann]

I have decided to not work on this and have deleted the branch.