[FIXED] Is it possible to downgrade ipykernel to 4.10.0 without downgrading to Spyder 3?

Issue

I am having some trouble with the Spyder IPython kernel crashing when I use the Tkinter backend. According to this bug report, it is a problem with the latest version of ipykernel. Additionally, when I use the qt5 backend it is very slow -- this is also a problem with version 5 of the ipykernel (I checked by downgrading ipykernel and using spyder 3, then updating ipykernel again and reproducing the behavior).

I would like to downgrade the ipykernel package to version 4.10.0 without downgrading from Spyder 4 to Spyder 3. Is this possible? When I use the conda command conda install ipykernel=4.10.0 it wants to automatically downgrade Spyder, but I wonder if there is a workaround.

Solution

This is not really an answer to the original question, but I created a bug report here, and it was discovered that the bug with qt4 and qt5 rendering matplotlib plots is specific to ipykernel version 5.3.3. To fix that particular problem, you only need to downgrade to version 5.3.2 using the command conda install ipykernel=5.3.2 which is compatible with IPython 4.

Additionally, it looks like a patch has been added to the Tkinter bug (see link in original question) that any user can add themselves. It appears that all of this should be fixed once ipykernel 5.3.4 gets added to the default conda channel.

Answered By - Lucas Myers