Actually ignore invalid log-axis limit setting #10950
Conversation
| def test_invalid_log_lims(): | ||
| # Check that invalid log scale limits are ignored | ||
| fig, ax = plt.subplots() | ||
| ax.scatter(range(0, 4), range(0, 4)) |
There was a problem hiding this comment.
Are you intentionally using data points that are not part of the valid log scale? I would have taken range(1, 3).
There was a problem hiding this comment.
No, but I don't think it matters really, as long as the limits don't change. The test fails on the master branch, so I think it's a meaningful test.
There was a problem hiding this comment.
It's definitively meaningful. I just had to put in a bit of extra thought what the actual limit before set_xlim would be. So it's slightly more complicated than need be.
Change it or leave it as you wish.
There was a problem hiding this comment.
Would this fix #10782? In that case using negative values explicitely might make sense for a test (testing for 0 and a negative value independently, I mean).
There was a problem hiding this comment.
No, it doesn't. I think that bug is a "scatter markers don't do autoscaling" bug.
lib/matplotlib/axes/_base.py
Outdated
| 'Attempted to set non-positive xlimits for log-scale axis; ' | ||
| 'invalid limits will be ignored.') | ||
| if self.get_xscale() == 'log': | ||
| if left <= 0.0: |
There was a problem hiding this comment.
actually 0 instead of 0.0 would be sufficient (also following occurrences). Not a merge blocker though, if you don't want to change it.
This PR makes sure the warning does what it says on the tin, which is not taking into account limits which are <= 0 on a log scaled axis.
Fixes #7733.