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
{{ message }}
Repository navigation
Update Display Data disappears when display and display.update are in the same cell #1710
This bug seems to be related to microsoft/vscode-python#10874 . Note that the consequence of this bug is that VSCode cannot be used effectively with the current version of the FastAI library.
Steps to cause the bug to occur
# %%
from time import sleep
from IPython.display import Markdown
# %%
dh = display(display_id=True)
# %%
dh.update(Markdown('`Display Update Fun 1`'))
sleep(3)
dh.update(Markdown('`This update remains`'))
# %%
dh = display(Markdown('`Display Update Fun 2`'), display_id=True)
dh.update(Markdown('`This update is supposed to remain`'))
sleep(3)
Actual behavior
Direct kernel connection
Python version:
3.7.3 (default, Apr 24 2020, 18:51:23) \n[Clang 11.0.3 (clang-1103.0.32.62)]
[1] from time import sleep...
[2] dh = display(display_id=True)
`This update remains`
[3] dh.update(Markdown('`Display Update Fun 1`'))...
[4] dh = display(Markdown('`Display Update Fun 2`'), display_id=True)...
`Display Update Fun 2`
Expected behavior
Direct kernel connection
Python version:
3.7.3 (default, Apr 24 2020, 18:51:23) \n[Clang 11.0.3 (clang-1103.0.32.62)]
[1] from time import sleep...
[2] dh = display(display_id=True)
`This update remains`
[3] dh.update(Markdown('`Display Update Fun 1`'))...
[4] dh = display(Markdown('`Display Update Fun 2`'), display_id=True)...
`This update is supposed to remain`
hsm thanks for the report, and sorry about the issue. I can confirm that something is off here. I repro the same issue and see that jupyter displays it correctly:
I also found the similar problem.
the proper behavior is an below.
but right after fine_tune completed, the result disappeared. when you close and reopen the .ipynb
the reslut (below) is there
The only workaround I found was to call clear_output() and then use DisplayHandle.display() again, but that's not a good option because it clears all the output.
Avram Lubkin (@avylove) Please could you install VS Code Insiders and try the new notebook UI and let us know if the problem has been resolved there?
Thanks
Avram Lubkin (@avylove) Sorry, you're hitting #5804. As a temporary workaround, you can replace the workbench.editorAssociations setting with the following:
The workaround wasn't working for me, but the new version did. There was some hanging with new interface at times, but this particular issue looks like it's resolved there.
I'm confused to why this was closed, particularly with no comment. While it seemed to work in the insiders version, that doesn't look like it's made it to general release yet. Please leave open until issue is deployed in a general release.
Activity
IanMatthewHuff commentedon Sep 1, 2020
hsm thanks for the report, and sorry about the issue. I can confirm that something is off here. I repro the same issue and see that jupyter displays it correctly:

emepetres commentedon Sep 10, 2020
I can confirm the issue. Below a couple of screenshots from the first notebook of the fastbook:
During training:

After training:

zhangjiekui commentedon Sep 22, 2020
I also found the similar problem.

the proper behavior is an below.
but right after fine_tune completed, the result disappeared. when you close and reopen the .ipynb
the reslut (below) is there
HeroadZ commentedon Sep 28, 2020
Same problem here. The result disappers when the training finished.
avylove commentedon May 8, 2021
Any movement on this?
The only workaround I found was to call
clear_output()and then useDisplayHandle.display()again, but that's not a good option because it clears all the output.DonJayamanne commentedon May 10, 2021
Avram Lubkin (@avylove) Please could you install VS Code Insiders and try the new notebook UI and let us know if the problem has been resolved there?
Thanks
avylove commentedon May 10, 2021
Having an issue with the Jupyter extension on Insiders.
extension activation failed TypeError: o.getConfiguration(...).get(...).find is not a functionOr were you referring to something else?
joyceerhl commentedon May 10, 2021
Avram Lubkin (@avylove) Sorry, you're hitting #5804. As a temporary workaround, you can replace the workbench.editorAssociations setting with the following:
which should allow the extension to activate.
avylove commentedon May 11, 2021
The workaround wasn't working for me, but the new version did. There was some hanging with new interface at times, but this particular issue looks like it's resolved there.
avylove commentedon Jun 6, 2021
I'm confused to why this was closed, particularly with no comment. While it seemed to work in the insiders version, that doesn't look like it's made it to general release yet. Please leave open until issue is deployed in a general release.