Improve testing by moving common code to testing module#1968
Improve testing by moving common code to testing module#1968
Conversation
|
Not ready to be merged but any feedback is welcome. @doutriaux1 @sankhesh @danlipsa @chaosphere2112 |
|
The core idea is to provide a regression testing module that provide api that is needed for most of the vcs regression testing. This work reduces the testing code to 20 - 60% in most of the cases. |
d1906cb to
9204ea6
Compare
|
Data: CDAT/uvcdat-testdata#132 |
c1cc6da to
73c3d4b
Compare
73c3d4b to
5dd0a94
Compare
|
@aashish24 I like this so far, I just gave it a quick glance but I like the concept. |
f9e5051 to
d9f6525
Compare
fbab187 to
a1f6064
Compare
|
thanks @doutriaux1. This branch is ready for review now. |
a1f6064 to
a2ce6ac
Compare
|
@danlipsa @doutriaux1 would it be possible to get it merged soon so that I don't need to keep upto date with master? |
|
ok let em build this on my mac an run the test suite on it. |
|
@aashish24 sorry my mac is very very slow.... still building |
|
@aashish24 test_vcs_monotonic_decreasing_yxvsx_default.png has some points along the line that where not present in the original image. Do you know why? |
|
@danlipsa I just saw test_vcs_monotonic_decreasing_yxvsx_default.png too. Let me have a look. |
|
@danlipsa I just looked. My change should not have caused it but I am looking for suggestions. |
|
@doutriaux1 vcs_click_info is failing because of dimension issues. I will push a fix. |
|
@aashish24 strange. I just run the test on my branch and it generates the (smaller) image without the dots. I would say to file an issue for this, if we think it is small enough to let it through. |
|
okay thanks @danlipsa I will file the issue. |
|
Ref: #1976 |
|
I am running the tests again, this time I think I should have it. It seems to me that in parallel build, I didn't get these errors (weird). |
|
should be all good now crossing my finger |
|
@danlipsa @doutriaux1 please try again. |
This is uvctest-data |
|
test_vcs_click_info is failing because it has to be in the foreground and width and height are not set. I tried to set geometry afterwards but looks like I found a bug. I will try to pass the geometry in the init call and see if that produces consistent behavior. |
|
@danlipsa I made the size of click tests to a fixed size. Can you try again please? |
|
@aashish24 Those pass now, but it seems that you changed something else. I have 114 tests that fail - the size of the generated image is different - unless I did something wrong. Are those passing on your machine? |
|
@aashish24 same here on mac, more failures now: https://open.cdash.org/viewTest.php?onlyfailed&buildid=4371882 |
21ded89 to
a142008
Compare
|
@danlipsa @doutriaux1 my bad, i relied on something from vcs which wasn't quite right. I will push a fix for vcs in a separate branch. In the mean time if you checkout the current SHA you should see only these tests failing: https://open.cdash.org/viewTest.php?onlyfailed&buildid=4372388 This should be the SHA: [chaudhary@einstein src.git (improve_testing_somemore)]$ |
|
@aashish24 Now it is good. These are the only failing tests. |
|
thanks @danlipsa finally 😌 @doutriaux1 okay to merge? |
|
sorry lots of fires this morning, let me run it on my mac just to be safe. |
|
@doutriaux1 looks like you have an older copy of regression.py. If you look at the code here; https://github.com/UV-CDAT/uvcdat/blob/9c121a20f2121a0fcca665123d50b27553f176a0/Packages/testing/regression.py you will see that init takes arguments. Can you make sure that you have latest code installed (and the baseline). |
|
@doutriaux1 Lets merge this in, otherwise nothing can get merged - this has lots of changes! Seems it is ready. |
|
failed on my mac yesterday, but I'm off (officially today) I trust you @danlipsa go ahead and merge |
No description provided.