position of text annotations looses unit information #4904
Conversation
after it was set. This fixes Text such that it will return the original data when asked and will only strip away the units when it is needed.
|
From git blame on this section of the code it looks like stripping the units has been the behavior since a43661e in from 2007. I agree this should go in, but could you please add an It might be worth making |
|
I can change '_get_unitless_position' to be public if that is what you want. I am not sure what you mean by add an 'api_changes' entry. |
|
See matplotlib/doc/api/api_changes/README On Mon, Aug 17, 2015, 3:55 PM James Evans notifications@github.com wrote:
|
|
He is referring to "doc/api/api_changes/". On Mon, Aug 17, 2015 at 3:55 PM, James Evans notifications@github.com
|
Renamed '_get_unitless_position' to 'get_unitless_position'
|
All of the test failures look like they were due to the network issues gh was having when they ran (all failed out on trying to hit a url for the style module or known transient failures). Kicked travis to restart ,will merge as soon as it comes back green. I am 👍 on this API change because it only effects people who are using units and the behavior they expect is what it has been changed to. @jrevans Could you follow this up with a test? I think just showing values round-trip with units is enough. |
API : Text.get_position now preserves units
A person could set the position using units, but when getting them back, the units would be stripped away. This made no sense since the unitized data was still present and the units were only being stripped away for use in internal methods. This fixes the behavior so the user will get back exactly what they set and the internal routines will still get the unitless data as needed.
This addresses an issue in #4897.