Closed marscher closed 5 years ago
I guess it's plt.annotate
because they changed the API. Maybe I can fix that.
On 16.11.18 10:38, Tim Hempel wrote:
I guess it's the
plt.annotate
because they changed the API. Maybe I can fix that.if I omit the arrowprops it succeeds.
Bug was reported to Matplotlib (https://github.com/matplotlib/matplotlib/issues/12820), for now resolved with #181 .
I sill get this bug, in the case of saving the figure as an image (format png, jpg, tiff). Saving it as a pdf works! The problem is the annotate function.
OK, interesting. We'll have to wait for the matplotlib developers to really fix this (it's reported, cf. post above). Good that you can save it to PDF, so I suggest you stick to this as a temporary solution (and convert to png/jpeg if necessary).
I sill get this bug, in the case of saving the figure as an image (format png, jpg, tiff). Saving it as a pdf works! The problem is the annotate function.
Maybe this helps to find the bug. If I add to the x-position a tiny distance of 0.000000000001, than is works:
axarr[0].annotate('', xy=(objectZValues[3]+0.000000000001, objectIntensities[3]), xytext=(objectZValues[3],-.05), bbox=dict(boxstyle="round", fc="w"), arrowprops = settings.arrowProps)
Yes, this is how we temporarily fixed this bug here. Are you sure you are posting to the right repository?
@thempel oh sorry, I was redirected here.
In nb08 this exception is raised during rendering the figure produced by plot_1D_histogram_trajectories. When I comment out the ax.annotate call, then the figure is rendered correctly. I assume this is triggered by a rigorous check, which has been added recently. I will report it upstream.
In the meantime we can simply downgrade matplotlib for test execution.