Jupyter: new "kernelspec" tag produces "jupyter_client.kernelspec.NoSuchKernel" with "jupyter nbconvert"

Environment data

• VS Code version: 1.55.0
• Jupyter Extension version: ms-toolsai.jupyter v2021.5.702919634
• Python Extension version: ms-python.python v2021.3.680753044
• OS and version: MacOSX Catalina 10.15.7
• Python: 3.8.8; Anaconda: 4.3.27
• Type of virtual environment used: conda
• Jupyter server running: Local

Expected behaviour

1. Save mynb.ipynb in VS Code
2. Convert notebook into html via terminal in the same env using jupyter nbconvert --to html mynb.ipynb --no-input --execute

Actual behaviour

Conversion produces error jupyter_client.kernelspec.NoSuchKernel: No such kernel named python388jvsc74a57bd040b90fc7ac2e82be620538bbf156a62f078051c1c11595b3240bdeeddbe7576e

Steps to reproduce:

1. mynb.ipynb:

{
 "metadata": {
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  },
  "orig_nbformat": 2,
  "kernelspec": {
   "name": "python388jvsc74a57bd040b90fc7ac2e82be620538bbf156a62f078051c1c11595b3240bdeeddbe7576e",
   "display_name": "Python 3.8.8 64-bit ('my_env38': conda)"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2,
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(\"test\")"
   ]
  }
 ]
}

1. jupyter nbconvert --to html mynb.ipynb --no-input --execute

The problem appeared after updating both VS Code and conda, though older notebooks can be converted without any problems in the same env.

The issue can be manually fixed if the “kernelspec” tag is removed:

,
  "kernelspec": {
   "name": "python388jvsc74a57bd040b90fc7ac2e82be620538bbf156a62f078051c1c11595b3240bdeeddbe7576e",
   "display_name": "Python 3.8.8 64-bit ('my_env38': conda)"
  }
 },

I’m using jupyter nbconvert because I have not found any ways to hide/remove cells with the built in exporter (the --no-input flag).

[rchiodo]

Do we need to do this same kernel name argument for our export?

[DonJayamanne]

@mihkelvaher I believe this has been resolved, we no longer store the generated name in the ipynb file. Please feel free to re-open this issue if the issue persists or please do create a new one and we’ll look into it.