You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When using the nbagg backend, the interactive widgets appear, but the plot itself does not, and the message [IPKernelApp] ERROR | No such comm: c795df2a293345a1ae3237c0f203d6d3
is printed to the terminal.
@takluyver@minrk Something changed between 2.3.1 and the master branch which has really broken nbagg.
We should get this figured out before ipython 3.0. Mpl have a 1.4.3 planed for ~end of the month anyway so if fixes are needed on the mpl we can get that out in parallel.
It looks like there has been a lot of work done on the logic of registering/unregistering comms, so I suspect something got broken in that process, but I don't understand the code well enough to be sure.
Yup, we'll dig into this to see if we can figure out what's gone wrong. Some code got cleaned up in creating comms, but I don't think any breakage was intentional, so there's a good chance there's a bug on our side.
Looks like it wasn't comm changes, but output changes that caused the problem. Opaque Promise failure messages just made it look like a comm failure, since it was an error in the comm_open callback. Should be fixedc by matplotlib/matplotlib#4002.
Activity
tacaswell commentedon Jan 2, 2015
@takluyver @minrk Something changed between 2.3.1 and the master branch which has really broken
nbagg.We should get this figured out before ipython 3.0. Mpl have a 1.4.3 planed for ~end of the month anyway so if fixes are needed on the mpl we can get that out in parallel.
tacaswell commentedon Jan 2, 2015
It looks like there has been a lot of work done on the logic of registering/unregistering comms, so I suspect something got broken in that process, but I don't understand the code well enough to be sure.
minrk commentedon Jan 2, 2015
Yup, we'll dig into this to see if we can figure out what's gone wrong. Some code got cleaned up in creating comms, but I don't think any breakage was intentional, so there's a good chance there's a bug on our side.
tacaswell commentedon Jan 12, 2015
@minrk Any update on this?
minrk commentedon Jan 12, 2015
Sorry, I'll try to dig into it some more.
minrk commentedon Jan 15, 2015
Looks like it wasn't comm changes, but output changes that caused the problem. Opaque Promise failure messages just made it look like a comm failure, since it was an error in the comm_open callback. Should be fixedc by matplotlib/matplotlib#4002.
Sorry it took so long for me to get around to it.
tacaswell commentedon Jan 15, 2015
@minrk fixed this on the mpl side, please close
minrk commentedon Jan 15, 2015
Thanks!