gtests/test_nopython_callprofiler.cpp timing is unstable

solvcon / modmesh

Toolkit for solving partial differential equations

BSD 3-Clause "New" or "Revised" License

29 stars 35 forks source link

gtests/test_nopython_callprofiler.cpp timing is unstable #347

Open yungyuc opened 1 month ago

yungyuc commented 1 month ago

See https://github.com/solvcon/modmesh/actions/runs/9378360062/job/25821452032#step:9:275 :

/Users/runner/work/modmesh/modmesh/gtests/test_nopython_callprofiler.cpp:175: Failure
Value of: diff_time(node1->data().total_time, uniqueTime1 + uniqueTime2)
  Actual: false
Expected: true

The test should use a more stable way to test for time.

tigercosmos commented 4 weeks ago

@yungyuc I checked all scheduled actions, no errors before, it's probably the PR issue. https://github.com/solvcon/modmesh/actions?page=3&query=event%3Aschedule

yungyuc commented 4 weeks ago

Right, it is a random failure. Please take a look at my quote and link in the issue description.

tigercosmos commented 4 weeks ago

@yungyuc No, it's not a random failure, it never happens in the past few months

yungyuc commented 4 weeks ago

I see what you meant. It does not happen with scheduled runs, but only happens with PR runs?

tigercosmos commented 4 weeks ago

@yungyuc yes, that's what I meant. The current code should be very stable in the master branch, so the PR owner need to check why it failed.

yungyuc commented 4 weeks ago

This is good finding. Thanks, @tigercosmos . Let me unassign this issue from you and move it out in progress. We should revisit when we find another occasion.