INVALID: Framework: PR build showing missing test executable but no build failure

[bartlettroscoe]

Bug Report

@trilinos/framework, @jhux2, @csiefer2, @e10harvey

Description

There are cases where the PR build results posted to CDash show a not run test due to a missing test executable but no build failure. For example, for the builds:

• PR-10808-test-rhel7_sems-clang-10.0.0-openmpi-1.10.1-serial_release-debug_shared_no-kokkos-arch_no-asan_no-complex_no-fpic_mpi_no-pt_no-rdc_no-uvm_deprecated-on_no-package-enables-810
• PR-10808-test-rhel7_sems-clang-10.0.0-openmpi-1.10.1-serial_release-debug_shared_no-kokkos-arch_no-asan_no-complex_no-fpic_mpi_no-pt_no-rdc_no-uvm_deprecated-on_no-package-enables-797
• PR-10808-test-rhel7_sems-clang-10.0.0-openmpi-1.10.1-serial_release-debug_shared_no-kokkos-arch_no-asan_no-complex_no-fpic_mpi_no-pt_no-rdc_no-uvm_deprecated-on_no-package-enables-792

showing:

...

for PR #10808, it showed the "Not Run" test ThyraEpetraAdapters_EpetraOperatorWrapper_UnitTests_MPI_4 with status Not Run link showing:

Unable to find required file: /scratch/trilinos/jenkins/ascic115/workspace/Trilinos_PR_clang-10.0.0/pull_request_test/packages/thyra/adapters/epetra/test/EpetraOperatorWrapper/ThyraEpetraAdapters_EpetraOperatorWrapper_UnitTests.exe

Well, how can that file be missing if there is no build error?

That suggests there is some type of framework problem here with how the build is being done, or how results are packaged up to send to CDash, or a problem with transmitting results to CDash, or a problem with how results are process and displayed on CDash.

I guarantee that is zero chance that I can reproduce this test as "Not Run" after running the "all" target without seeing a build error.

Steps to Reproduce

The fact that this strange state was reported in all three PR iterations for the same build suggests that this is a repeatable bug (but only in the autotester).

[bartlettroscoe]

FYI: This is likely due to using an old version of CMake before global error reporting was implemented. See #10893 (comment)

Closing as invalid.