[TST] Added test_axlines in test_datetime.py #27326
Conversation
There was a problem hiding this comment.
Thank you for opening your first PR into Matplotlib!
If you have not heard from us in a week or so, please leave a new comment below and that should bring it to our attention. Most of our reviewers are volunteers and sometimes things fall through the cracks.
You can also join us on gitter for real-time discussion.
For details on testing, writing docs, and our review process, please see the developer guide
We strive to be a welcoming and open project. Please follow our Code of Conduct.
There was a problem hiding this comment.
This has revealed that axline does not support units currently, but I think it could with relative ease, see my other comment for more details
Since it does not support units currently, we are not actually testing the intended code paths here, so it shouldn't be merged as is.
| values = list(range(1, 32)) | ||
|
|
||
| ax1.plot(dates, values, 'ro') | ||
| ax1.axline((date2num(dates[0]), values[0]), (date2num(dates[-1]), values[-1]), |
There was a problem hiding this comment.
We specifically do not want to use date2num in these test cases. the fact that it is erroring when handed datetime objects is precisely the kind of thing we wanted to find out by implementing these tests.
The following patch gets most of the way to fixing it such that you can actually pass datetime objects (though I think it may need some consideration regarding what to do if a non-data transform is provided, or whether "slope" is something that can be meaningfully provided with unitful data):
diff --git a/lib/matplotlib/axes/_axes.py b/lib/matplotlib/axes/_axes.py
index 9199f1cafa..524c1ae06a 100644
--- a/lib/matplotlib/axes/_axes.py
+++ b/lib/matplotlib/axes/_axes.py
@@ -919,6 +919,9 @@ class Axes(_AxesBase):
self.get_yscale() != 'linear'):
raise TypeError("'slope' cannot be used with non-linear scales")
+ xy1 = tuple(self._process_unit_info([("x", xy1[0]), ("y", xy1[1])]))
+ if xy2 is not None:
+ xy2 = tuple(self._process_unit_info([("x", xy2[0]), ("y", xy2[1])]))
datalim = [xy1] if xy2 is None else [xy1, xy2]
if "transform" in kwargs:
# if a transform is passed (i.e. line points not in data space),There was a problem hiding this comment.
I did the minimal change here to get the point-point version working with the default transform (data space). I do think that the final version would need at least some more.
- passing
kwargsin - Maybe pushing the unit transforming down into
mlines.AxLine? (trying to trace how baseLine2Ddoes it, if that does it in the class, I'd probably follow suit forAxLine, but not super clear on that, actually...AxLineinherits fromLine2D) - some more thought about
slopemay be good... it is already disallowed with non-linear scales, may be best to just disallow it with unitful data or declare it to be in explicitly "unitless data space" (currently it states that it is in 'data coordinates'). I think it may actually work out for at least the case of having the same units on both axes (i.e. the slope is unitless then) or possibly also with full unit libraries, though possibly would require pushing units into theAxLineclass for it to work.
PR summary
This PR adds a smoke test for Axes.axlines in lib/matplotlib/tests/test_datetime.py. This PR closes one of the items listed in the issue #26864.
Smoke test Plot:
PR checklist