Fix bar3d bug with matching color string and array x lengths #15826
Conversation
|
Why did these checks fail? |
|
The failure on the nightly job was due to a change in numpy that broke our test suite (they are now warning someplace we did not expect and we fail the test suite on warnings) which is not your problem. On the master branch we now mark the nightly job as an "allowed failure" so the next time you push a commit it it won't fail. Lines changed in the coverage report are clearly unrelated to this PR, I suspect it is a glitch in the coverage reports. Can you add a test for this? I don't think we need an image test. It is probably enough to test that it does not fail and assert that the color of the returned bar objects in orange. Thank you for you work on this! 🎆 |
I was thinking of doing that by comparing the RGBA values in the |
|
The original issue crashes, so a test like that checks both that it has the expected result and doesn't crash. I don't think there any need to find some better kind of test. |
8df22f0 to
7817680
Compare
|
I pushed a simple smoke test, so that we can merge this for 3.3. |
7817680 to
680e839
Compare
PR Summary
Closes #15815
The bug is caused by the line
len(color) == len(x), sincelen(color)is the length of string rather than the actual number of colors.Converting color to an RBG array before comparison ensures that
len(color)actually represents the number of colors.PR Checklist