Make the BuildBot green on master branch (Bug #2597)
Description
As we discussed, master branch should be always green. But that is not the case right now, since we have 12 failing tests. I suggest to come to a stable state and later register every time when a pull request passes the pre-commit testing, but later breaks the major BuildBot. This will help to test patches more strictly.
Vadim, how will we work on the current list of issues?
- Calib3d_FindEssentialMat.accuracy
- Calib3d_SolvePnPRansac.accuracy
- Core_Invert.small
- Features2d_FLANN_Auto.regression
- Highgui_Video_parallel_writers_and_readers.accuracy
- opencv_test_imgproc
- GPU_Denoising/FastNonLocalMeans.Regression/0, where GetParam
- GPU_ImgProc/HoughCircles.Accuracy/0, where GetParam
- GPU_ObjDetect/HOG.Detect/0, where GetParam
- GPU_SoftCascade/SCascadeTestAll.detect/0, where GetParam
- GPU_SoftCascade/SCascadeTestAll.detectOnIntegral/0, where GetParam
- GPU_SoftCascade/SCascadeTestAll.detectStream/0, where GetParam
These tests failed on precommit_windows
builder, so I think there are no real issues.
- SCascade.detect
- SCascade.detectNoRoi
- SCascade.detectRoi
- SCascade.detectSeparate
- SCascade.readCascade
Related issues
related to Bug #2773: investigate test failures on Windows buildbot builders | Cancelled | 2013-02-06 |
Associated revisions
fixed Video_calcOpticalFlowDual_TVL1 test (Bug #2597)
uses RMSE instead of bitwise equivalence
Merge pull request #2597 from ilya-lavrenov:ipp_countnonzero
History
Updated by Vadim Pisarevsky over 12 years ago
Core_Invert.small test should pass after merging 2.4 into master.
I will look at the other Core_* and Highgui_* tests.
SolvePnPRansac needs to be fixed, both the algorithm and the test.
Optical Flow TVL1 - need to ask the author (Alexey?) to look at it.
ML tests fail only on android_arm, so I leave fixing those tests to the tegra team.
On GPU tests - need to.
In other words, I will fix the tests I can fix and then reassign the ticket to the next relevant person.
Updated by Kirill Kornyakov over 12 years ago
OK, now I see that some tests are failing on Android only. It wasn't clear to me before...
Thank you for readiness to work on this! I've talked to Anatoly, and he promised to look on failures in 2 weeks (they are busy with some algorithm). We'll check all the Android issues, but they are looking strange, since we don't see such failures in our test environment. We'll check them anyway.
Updated by Andrey Kamaev over 12 years ago
Core_Invert.small test should pass after merging 2.4 into master.
It was merged on Friday.
Optical Flow TVL1 - need to ask the author (Alexey?) to look at it.
I think it was Vlad.
Updated by Kirill Kornyakov about 12 years ago
Updated description for today's state: Jan 10, 2013.
Marina, could you comment on SCascade issues, both GPU and precommit_windows?
- Description changed from As we discussed, _master_ branch should be always green. But that is not the ... to As we discussed, _master_ branch should be always green. But that is not the ... More
Updated by Kirill Kornyakov about 12 years ago
Here is ticket for the Report page.
- Affected version set to branch 'master' (2.4.9)
- Target version deleted (
2.4.4)
Updated by Vadim Pisarevsky about 12 years ago
- Assignee deleted (
Vadim Pisarevsky)
Updated by Kirill Kornyakov about 12 years ago
I would say that we've almost succeed with this task. Of course there will be other test failures, but we'll work on them on during daily work and release preparation.
- Target version set to 2.4.4
- Affected version changed from branch 'master' (2.4.9) to branch '2.4'
- Status changed from Open to Done