Bug summary

When using broken_barh with a categorical (vertical) axis, there is no obvious way to specify the height of the bars. Unlike regular barh, float values do not work in the second argument position to specify heights.

You can hack it by using other categorical values (see example below), but this only works for integral heights and is generally yucky.

Code for reproduction

import matplotlib.pyplot as plt

fig, ax = plt.subplots()

ax.broken_barh([(0, 5-0), (10, 13-10)], ('car engine', 1))
ax.broken_barh([(3, 6-3), ], ('jackhammer', 1))
ax.broken_barh([(1, 4-1), ], ('children playing', 1))

Above does not work. Substituting a categorical value (which maps to 1) does work:

import matplotlib.pyplot as plt

fig, ax = plt.subplots()

# 'jackhammer' is implicitly height=1 in this example
ax.broken_barh([(0, 5-0), (10, 13-10)], ('car engine', 'jackhammer'))
ax.broken_barh([(3, 6-3), ], ('jackhammer', 'jackhammer'))
ax.broken_barh([(1, 4-1), ], ('children playing', 'jackhammer'))

Actual outcome

ConversionError: Failed to convert value(s) to axis units: ('car engine', 1)

Expected outcome

It should be possible to specify heights by numerical values as well as / instead of categorical positions.

Additional information

For broader context, the full description of my use case is at https://discourse.matplotlib.org/t/advice-for-using-broken-bar-collections-with-categorical-axes/23171?u=brian.mcfee

Best guess for fix: set this up however it was done over in barh 🤷

Probably this also affects broken_bar (vertical) as well.

Operating system

No response

Matplotlib Version

3.6.3

Matplotlib Backend

No response

Python version

No response

Jupyter version

No response

Installation

conda