Bug report

Bug summary

Code for reproduction

frommatplotlibimportpyplotaspltfig,axs=plt.subplots(2)axs[0].bar("thing",1,data={"thing":1})axs[1].bar("other",1,data={"thing":1})plt.show()

Actual outcome

The first plot's x-value ("thing") is interpreted as a lookup into the data dict and replaced by 1.
The second plot's x-value does not appear in the data dict and is interpreted as a categorical.

Expected outcome

Not sure what the best option is, but silently changing the interpretation of input based on whether it is present or not in another dict seems a bit finnicky. I have proposed in other places to not allow scalar categoricals at all (always need to be passed in a container -- list, array, dataframe, etc.), which also solves the inconsistency thatplot(1, "x") currently specs a marker whereasplot("x", 1) treats"x" as a categorical.

Unlike other categorical issues I don't actually think this is release critical per se, but it would still be nice to get the behavior clarified/simplified...

Matplotlib version

• Operating system:
• Matplotlib version: 2.1
• Matplotlib backend (print(matplotlib.get_backend())):
• Python version:
• Jupyter version (if applicable):
• Other libraries: