Make code match comment in sankey. #11974
Conversation
This makes Sankey.add no longer reject a labels that's a *single* string, consistently with what the docstring states and what the comment just after implies. Previously, it would only accept a single string if its length is equal to the number of flows, in which case it would use one character of the string for each flow. Alternatively, we could also change the docstring and delete the comment, but I think this behavior is more intuitive? (Not that I ever use Sankey, though it looks neat...)
|
👍 but could do with a test? |
|
Well, sankey tests are... sparse (https://github.com/matplotlib/matplotlib/blob/master/lib/matplotlib/tests/test_sankey.py) and I'm not volunteering to fix that :) |
|
This has broken the first two plots in the Sankey class example and the Rankine power cycle example. I think you've misdiagnosed the problem here. It appears that it's a Python 2 vs. 3 bug. Python 2 What it was originally supposed to check is that if you passed a list (or tuple), that it was the right length. Passing a single string was to be repeated for all entries. |
|
Yes, I just realized this while testing the other sankey PRs... working on it. |
This makes Sankey.add no longer reject a labels that's a single
string, consistently with what the docstring states and what the comment
just after implies.
Previously, it would only accept a single string if its length is equal
to the number of flows, in which case it would use one character
of the string for each flow.
Alternatively, we could also change the docstring and delete the comment,
but I think this behavior is more intuitive? (Not that I ever use
Sankey, though it looks neat...)
Noted in #11959, dates all the way back since this class was added.
PR Summary
PR Checklist