Jupyter notebook -- disappearing cells?

I had a Jupyter notebook running in my browser, and suddenly I was told that there was a problem with connection to the server. It seemed like Julia had crashed…

When I open notebooks after that, some cells don’t show in the browser, but are indicated by horizontal lines:
image

Apparently, the code is still there. When I open the notebook in an ASCII editor, I observe that code cells that still are visible, have the following structure:

"cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "x0 = [28., 28., 28., 14.,18.,22.]\n",
    "tspan = (0.0,583*60.)\n",
...

(observe "execution_count": 12,), while cells that are hidden look like:

"cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Model vs data\n",
    "idx = 1:10:584\n",

(observe "execution_count": null,).

Questions:

  • Is the problem that the execution count has been set to null?
  • Can I manually change this to some number? Do the execution_count number has to be an integer in sequence, starting from the top?

is it this (Cells not rendering or loading and code seems lost in jupyterlab notebook · Issue #10840 · jupyterlab/jupyterlab · GitHub)? If so, upgrade.

1 Like

I’ll check. I did Pkg.update(), but perhaps more is needed.

Yes, I used jupyterlab(), and the the notebook opens without problem in notebook().

OK… I did:

using Conda
Conda.rm("jupyterlab")

and then:

using IJulia
jupyterlab()

leading to a question whether I want to install jupyterlab. I answered “yes”.

The problem still persists.

sorry, I have no other ideas

1 Like

OK – I removed everything relating to Julia, Conda, Matplotlib, Jupyter, etc., and upgraded to Julia v1.6.3 + reinstalled packages.

Now, it seems like I can open the notebook using jupyterlab()… BUT after running for a while, I keep getting the following message:

I’m on Windows 11, and I have a feeling Microsoft updated Microsoft 365 on my computer yesterday, and possibly browser (chromium) Edge.

  • Could a bug in the Julia - Edge interaction have been introduced? [I’ll try to use Firefox as browser – unfortunately, in Windows 11 it is more complicated to change browser…]

I have been experiencing the same issue for about 3 days. I noticed that whenever I do not accept updates and postpone them for a future time, Windows (10 in my case) generates troubles. After installing updates the issues disappear.

In addition to updating to Windows 11 + possibly an update in Edge/Windows 365 on my computers, I have also been asked to use Conda to update PyPlot.