PLT: plot('line') or plot('area') produces wrong xlim in xaxis in 0.25.0 #27993
Conversation
charlesdong1991
changed the title
charlesdong1991
changed the title
Codecov Report
@@ Coverage Diff @@
## master #27993 +/- ##
==========================================
- Coverage 93.05% 93.04% -0.01%
==========================================
Files 189 189
Lines 50058 50042 -16
==========================================
- Hits 46581 46564 -17
- Misses 3477 3478 +1
Continue to review full report at Codecov.
|
| @@ -419,6 +419,7 @@ def test_get_finder(self): | |||
| assert conv.get_finder("A") == conv._annual_finder | |||
| assert conv.get_finder("W") == conv._daily_finder | |||
|
|
|||
| @pytest.mark.xfail # I am not sure if this test is correct | |||
There was a problem hiding this comment.
Can you expand on this? Perhaps open a new issue with the description of the bug, and reference it.
There was a problem hiding this comment.
Can you post images of the output for one of the failing tests, before and after your changes?
There was a problem hiding this comment.
@TomAugspurger yes, i feel the best way is to xfail the tests here and open up a new issue and do a follow-up PR to fix the tests, Because the main purpose is to test if the finder is being correctly used or not. And I honestly do not understand especially the second one in which there is a vmin + 0.9, I assume that's because of this wrong xaxis output, so has to manually add a constant to match the behavior.
There was a problem hiding this comment.
After my change: (this is also aligned with the behaviour on matplotlib)
Before my change:(i feel this is a bit wrong, because plot should start from 1987Q2, while due to this error, it starts from Q3 IIUC)
There was a problem hiding this comment.
Thanks... I think I agree with your assessment.
Can you open a new issue to discuss removing / updating those tests? And reference that in the skip / xfails?
There was a problem hiding this comment.
sure, i was thinking of doing it, but then i thought this PR would become a blocker for the issue if it is not merged because the test has to use the output of this PR. Probably I just overthink it ^^ thanks, will open and reference it. @TomAugspurger
|
Can you move the release note to 1.0.0.rst? |
|
Thanks @charlesdong1991! |
…5.0 (pandas-dev#27993) * Fix issue 27686
Due to previous PRs, there is an issue with xlim wrongly plotted for lines. I digged into database a bit, and found it also affects
plot(kind='area')because they share the sameLinePlot. And this issue is produced in both DataFrame and Series. Now looks like the issue is gone:Duplicated closed issue 27796 is solved as well:
Also looks like issue in #25160 is gone as well.
The same to #24784 looks like issue is also gone
black pandasgit diff upstream/master -u -- "*.py" | flake8 --diff