For example, see nightlies here: https://cdash.mitk.org/index.php?project=MITK&date=2020-11-14
Description
Description
Revisions and Commits
Revisions and Commits
rMITK MITK | |||
Restricted Differential Revision | rMITKd76d2dc174b0 Reduce number of labels in test label set | ||
Restricted Differential Revision | rMITKb72e8c02f3a3 Reduce test image dimensions | ||
Restricted Differential Revision | rMITK28f981c7cd55 Fix comparison between signed and unsigned integers | ||
Restricted Differential Revision | rMITK1198ee0afe28 Reduce execution times of longest tests |
Related Objects
Related Objects
- Mentioned In
- T27977: 2020 Week 47 (Mid November)
Event Timeline
Comment Actions
Could it be that the test time out time is 600 sec? Because in the last month all tests runs of LabelSetTest with worked (as long as the duration was below 600 sec). All failed tests took longer then 600 sec.
For what ever reason under win the test is relativaly fast ~150-200 sec. Under linux/mac it is lurking arround 590 sec +- 25...
Comment Actions
Yes, eventually the test times out. But even if it would "fix" the test to increase this value, individual unit tests should aim to last not longer than around a second, at maximum a few seconds.
Comment Actions
1.5 reasons:
- mitk::LabelSet::AddLabel() takes a pretty long time to complete while it fiddles around with an std::map
- Not only does every test unnecessarily initialize a label set with 200 labels, but one test adds another 65535 labels to test the maximum allowed number of labels