Windows Interface test failures

[andrjohns]

While running a workflow to check the interface tests across the various Github runners (for the clang crash), the windows tests have a failure (for both x86_64 and arm64):

Running main() from stan/lib/stan_math/lib/benchmark_1.5.1/googletest/googletest/src/gtest_main.cc
[==========] Running 6 tests from 1 test suite.
[----------] Global test environment set-up.
[----------] 6 tests from CommandDiagnose
[ RUN      ] CommandDiagnose.corr_gauss
[       OK ] CommandDiagnose.corr_gauss (160 ms)
[ RUN      ] CommandDiagnose.corr_gauss_depth8
[       OK ] CommandDiagnose.corr_gauss_depth8 (157 ms)
[ RUN      ] CommandDiagnose.corr_gauss_depth15
[       OK ] CommandDiagnose.corr_gauss_depth15 (159 ms)
[ RUN      ] CommandDiagnose.eight_schools
[       OK ] CommandDiagnose.eight_schools (31 ms)
[ RUN      ] CommandDiagnose.mix
src/test/interface/diagnose_test.cpp:80: Failure
Value of: out.hasError
  Actual: true
Expected: false
"bin\diagnose src\test\interface\example_output\mix_output.*" quit with an error
[  FAILED  ] CommandDiagnose.mix (10 ms)
[ RUN      ] CommandDiagnose.divergences
src/test/interface/diagnose_test.cpp:95: Failure
Value of: out.hasError
  Actual: true
Expected: false
"bin\diagnose src\test\interface\example_output\div_output*.csv" quit with an error
[  FAILED  ] CommandDiagnose.divergences (10 ms)
[----------] 6 tests from CommandDiagnose (527 ms total)

[----------] Global test environment tear-down
[==========] 6 tests from 1 test suite ran. (527 ms total)
[  PASSED  ] 4 tests.
[  FAILED  ] 2 tests, listed below:
[  FAILED  ] CommandDiagnose.mix
[  FAILED  ] CommandDiagnose.divergences

 2 FAILED TESTS
test\interface\diagnose_test.exe --gtest_output="xml:test/interface/diagnose_test.xml" failed
exit now (06/17/25 14:26:22 Coordinated Universal Time)

[WardBrian]

These are passing on Jenkins. My guess is something about the way we read in test files isn't super portable

[andrjohns]

These are passing on Jenkins. My guess is something about the way we read in test files isn't super portable

Ah gotcha, will see if I can reproduce/debug locally then