It's in API changes; this is not a bug, but a correction of less-desired behaviour. The line width is now actually consistent with the width you requested.

Thank you for your answer.
but,
these are not good news for me.

oh, you are so fast
I will try it

Please allow me a statement:
In mpl 2.0.0 very thin and sharp lines for the V and H lines are available for the cursor. But not so in mpl 2.0.1. This is a step back. I would appreciate it if the developers rethink that.
Thank you

The API change is due to #8032 (the relevant commit is 276e1e9). Basically the former scaling scheme was dashes = unscaled_dashes * max(2, linewidth).

In your case, what about something like

import numpy as np
import matplotlib.pyplot as plt
from matplotlib.widgets import Cursor

# Manually rescale the desired dash pattern
# NB: plt.rcParams['lines.dashdot_pattern'] = [6.4, 1.6, 1.0, 1.6] but
# one could simply define our own on-off ink sequence (in points)
my_dashdot = [2.0*val for val in plt.rcParams['lines.dashdot_pattern']]
cursor_props = dict(linewidth=0.5, dashes=my_dashdot, color='#660099')

plt.close('all')
fig, ax = plt.subplots()

t = np.arange(0.0, 3.0, 0.01)
s = np.sin(2*np.pi*t)
ax.plot(t, s)

line_cursor = Cursor(ax, **cursor_props)

plt.show()

Edit: English…

Thank you for the example.
It looks better.
I still need to figure out how to do it for matplotlib.figure
...
I replaced plt.rcParams with matplotlib.rcParams and it seems to work.
I hope I'm on the right track.

many thanks for the support

I am pretty sure matplotlib.pyplot.rcParams is simply a convenient shortcut for matplotlib.rcParams.

BTW, using the dash-dot pattern from rcParams was just a mean of showing that you could manually scale one of the default patterns. But you could also perfectly use your own dash sequence if you are not satisfied with it. See the Line2D docstring for a comprehensive description of the API, or simply this example to get a quick glance at the syntax. In a nutshell, a dash style is an even-length sequence of on/off ink (in points).

PS: a similar way to customize the dashes may be to define a custom line style, which can be seen as a custom dash style prepended with an “offset” (usually set to zero...), following the pattern (offset, on-off-dash-seq). See for example here.

Thanks for your help, I'm not a professional developer even though I've created a complex Matplotlib application
I'm glad to have the right cursor lines again

ANT-fit plot

That's a nice looking application :).