First, this can be made even simpler:

from decimal import Decimal
import matplotlib.pyplot as plt
fig, ax = plt.subplots()
x = [Decimal(x) for x in range(10)]
y = [Decimal(x) for x in range(10)]
bar1 = ax.bar(x, y)

yields:

 File "/Users/jklymak/matplotlib/lib/matplotlib/axes/_axes.py", line 2250, in bar
    left = x - width / 2
TypeError: unsupported operand type(s) for -: 'decimal.Decimal' and 'float'

The problem is that we have to do some math on x to make bars, and we can't do mixed math on Decimal. (this is unlike other functions like plot or scatter, which don't seem to mind iterating over the values in x, be they decimal or numpy...)

We can't just force it to float, because other values of x need not be floats (i.e. datetime, categorical).

A possible solution is to write a unit converter for decimal and add to the default converters. Of course this would just make everything np.float, in which case, why not just cast it to float at the user level.

This also affects datetime.time values on the x axis. Here's an example that gives a similar error:

import datetime
import matplotlib.pyplot as plt
xs = [datetime.time(x) for x in range(10)]
ys = list(range(10))
plt.bar(xs, ys)

yields

/usr/local/envs/py3env/lib/python3.5/site-packages/matplotlib/axes/_axes.py in bar(self, *args, **kwargs)
   2064         if align == 'center':
   2065             if orientation == 'vertical':
-> 2066                 left = x - width / 2
   2067                 bottom = y
   2068             elif orientation == 'horizontal':

TypeError: unsupported operand type(s) for /: 'list' and 'int'

But it works fine using plt.plot.

Matplotlib version 2.1.2

@Stigjb Your issue is different. You are trying to plot time values which matplotlib in general does not support. For a workaround, use datetime instead of time values.

In case you want matplotlib to add support for time values natively, I would suggest you open a new issue with this feature request.

I am a student and is new to Matplotlib. For the problem about Decimal and float type error, I think it is because the height and width in bar(...) function in _axes.py (based on my current understanding) are always of type float (or integer). Further more, when creating patches, the width and height may be used in constructors. Here is my thought about how to (kind of) fix it:
Right before we compute alignment, we add a check:

if orientation=='vertical' and x.size > 0 and isInstance(x[0], Decimal):
     width = np.array([Decimal(itm) for itm in width ])
elif orientation=='horizontal' and y.size > 0 and isInstance(y[0], Decimal):
     height = np.array([Decimal(itm) for itm in height])

# Insert above code right before the following code
if align == 'center':
    if orientation == 'vertical':
        try:
             left = x - width / 2
# the rest code unchanged

Here are some bar plots I can produced with Decimal x and y after the above change:

Since I am quite new to Matplotlib, I am wondering whether this can be a fix and whether it has any side effect.
Please do pardon my beginner knowledge.