Duplicated tick label of colorbar

[zxdawn]

Description

Sometimes there're duplicated ticks shown on the colorbar.

Steps to reproduce

fig, axs = pplt.subplots()
import numpy as np

Z = np.random.rand(5, 5)*3
m = axs.pcolormesh(Z, cmap='viridis', levels=256, vmin=0, vmax=3)
axs.colorbar(m, ticks=0.25)

Expected behavior: [What you expected to happen]
The label of 2 tick shouldn't be overlayed with 2.25

Actual behavior: [What actually happened]

Proplot version

matplotlib: 3.5.0
proplot: 0.9.5.post78

[lukelbd]

Sorry, I fixed an issue related to #303 (where outer colorbars are getting drawn twice) and thought that solved the problem. But, the problem is still present. I think it's actually related to the DiscreteLocator used with "discrete" levels -- if you use discrete=False instead of levels=256 the issue goes away (see this section of the docs -- note that in general, if you just want "smooth" levels, using discrete=False will give a faster rendering time than modifying levels).

[lukelbd]

After further investigation, it seems this is unrelated to DiscreteLocator -- it's a matplotlib 3.5+ bug related to "discrete" normalizers like matplotlib's BoundaryNorm and proplot's DiscreteNorm. Version 3.5 completely overhauls the colorbar ticking internals so this isn't too surprising.

Here's an example with matplotlib only that reproduces the issue (not sure how to fix the messed up color scale -- norm.vmax = 3 does nothing and passing vmax=3 to pcolormesh raises an error -- but that's not relevant). This problem also seems to be independent from the number of "boundaries". If you have time feel free to start a matplotlib issue with this example.

import numpy as np
import matplotlib.pyplot as plt
import matplotlib.colors as mcolors
import matplotlib.ticker as mticker
fig, ax = plt.subplots()
edges = np.linspace(0, 3, 13)
norm = mcolors.BoundaryNorm(edges, 1000)
Z = np.random.rand(5, 5) * 3
m = ax.pcolormesh(Z, cmap='viridis', norm=norm)
cb = fig.colorbar(m, ticks=mticker.MultipleLocator(0.25))
fig.show()

[zxdawn]

Thanks for the debug. Submitted the issue to matplotlib.