test results dependent upon previous run artifacts

[zenspider]

Runs should generate any data they need. It appears this is not the case. In the example below, the first test (using latest released minitest) passes fine. The second run (using my private minitest changes) happens to pass because of the first run. The third run fails because of the side effects of the second run.

% ruby -S minitest && ruby -I ~/Links/MT/lib -S minitest && ruby -I ~/Links/MT/lib -S minitest
Run options: --seed 48132

# Running:

.........................

Finished in 0.663076s, 37.7031 runs/s, 73.8980 assertions/s.

25 runs, 49 assertions, 0 failures, 0 errors, 0 skips

[Minitest::CI] Generating test report in JUnit XML format...
Run options: --seed 45501

# Running:

.........................

Finished in 0.652496s, 38.3144 runs/s, 75.0962 assertions/s.

25 runs, 49 assertions, 0 failures, 0 errors, 0 skips

[Minitest::CI] Generating test report in JUnit XML format...
Run options: --seed 61515

# Running:

....................F....

Finished in 0.673684s, 37.1094 runs/s, 72.7344 assertions/s.

  1) Failure:
TestMinitest::TestCiPlugin#test_testsuite_sets_timestamp [/Users/ryan/wtf/minitest-ci/test/minitest/test_ci.rb:95]:
Expected false to be truthy.

25 runs, 49 assertions, 1 failures, 0 errors, 0 skips

[Minitest::CI] Generating test report in JUnit XML format...

I don't have a fix for this.