VS Code: Failed to start the Kernel

[egg-juxt]

Hi, my tslab kernel is not starting when inside the VS Code Jupyter Extension I select the tslab TypeScript kernel.

Not sure if it is a problem with the extension, because it seems the JSON config file doesn't get created, so tslab is not able to open it. But I have tested with the Deno kernel and it does start.

From jupyter lab it does work.

Visual Studio Code (1.91.1, undefined, desktop)
Jupyter Extension Version: 2024.6.0.
Python Extension Version: 2024.10.0.
Pylance Extension Version: 2024.7.1.
Platform: linux (x64).
Workspace folder ~/src/temp/node-xtdb, Home = /home/egg
14:28:12.734 [debug] Start refreshing Kernel Picker (1722342492734)
14:28:12.852 [debug] KernelProvider switched kernel to id = .tslab./tslab./.tslab#kernel#--config-path#{connection_file}
14:28:12.852 [debug] start the kernel, options.disableUI=true for ~/src/temp/node-xtdb/src/sample2.ipynb
14:28:12.865 [trace] Registering commtarget jupyter.widget
14:28:12.868 [debug] Controller selection change completed
14:28:12.868 [trace] Version of IPyWidgets 7 determined after 0.001s
14:28:12.912 [info] Starting Kernel (TypeScript) for '~/src/temp/node-xtdb/src/sample2.ipynb' (disableUI=true)
14:28:12.913 [trace] Creating raw notebook for resource '~/src/temp/node-xtdb/src/sample2.ipynb'
14:28:12.981 [trace] Search for KernelSpecs in Interpreter /bin/python
14:28:12.982 [trace] Search for KernelSpecs in Interpreter /sbin/python
14:28:12.982 [trace] Search for KernelSpecs in Interpreter /usr/bin/python
14:28:12.983 [trace] Search for KernelSpecs in Interpreter /usr/sbin/python
14:28:12.983 [trace] Kernel Spec for 'Python 3 (ipykernel)' (/usr/share/jupyter/kernels/python3/kernel.json) is hidden. (isDefaultKernelSpec = true, language = python, registrationInfo = undefined)
14:28:12.985 [debug] Get Custom Env Variables, Class name = Sm, completed in 250ms, has a truthy return value, Arg 1: undefined, Arg 2: "RunPythonCode"
14:28:12.985 [debug] Jupyter Paths /kernels: 
14:28:12.985 [debug] Kernel Spec Root Paths, /usr/share/jupyter/kernels, /usr/local/share/jupyter/kernels, ~/.local/share/jupyter/kernels
14:28:13.290 [debug] End refreshing Kernel Picker (1722342492734)
14:28:13.292 [debug] Launching kernel .tslab./tslab./.tslab#kernel#--config-path#{connection_file} for ~/src/temp/node-xtdb/src/sample2.ipynb in //home/~/src/temp/node-xtdb/src with ports 9001, 9000, 9004, 9002, 9003
14:28:13.292 [info] Launching Raw Kernel TypeScript # tslab
14:28:13.292 [debug] Create ProcessService, Class name = Im (started execution), Arg 1: "~/src/temp/node-xtdb/src/sample2.ipynb", Arg 2: {"a":false,"b":{"z":3,"w":{"i":-1,"end":0},"u":[{"id":1034},{"id":1035},null,{"id":1038}]}}
14:28:13.298 [debug] Get Custom Env Variables, Class name = Sm, completed in 6ms, has a truthy return value, Arg 1: "~/src/temp/node-xtdb/src/sample2.ipynb", Arg 2: "RunNonPythonCode", Arg 3: {"a":false,"b":{"z":3,"w":{"i":-1,"end":0},"u":[{"id":1034},{"id":1035},null,{"id":1038}]}}
14:28:13.298 [debug] Create ProcessService, Class name = Im, completed in 6ms, has a truthy return value, Arg 1: "~/src/temp/node-xtdb/src/sample2.ipynb", Arg 2: {"a":false,"b":{"z":3,"w":{"i":-1,"end":0},"u":[{"id":1034},{"id":1035},null,{"id":1038}]}}
14:28:13.305 [info] Process Execution: /tslab kernel --config-path /home/~/.local/share/jupyter/runtime/kernel-v2-29663rdSmjgt0v06u.json
    > cwd: //home/~/src/temp/node-xtdb/src
14:28:13.305 [debug] Kernel process undefined.
14:28:13.308 [error] Kernel died [Error: spawn tslab ENOENT
    at ChildProcess._handle.onexit (node:internal/child_process:286:19)
    at onErrorNT (node:internal/child_process:484:16)
    at process.processTicksAndRejections (node:internal/process/task_queues:82:21)] {
  errno: -2,
  code: 'ENOENT',
  syscall: 'spawn tslab',
  path: 'tslab',
  spawnargs: [
    'kernel',
    '--config-path',
    '/home/~/.local/share/jupyter/runtime/kernel-v2-29663rdSmjgt0v06u.json'
  ]
} 
14:28:13.309 [error] Disposing kernel process due to an error [Error: spawn tslab ENOENT
    at ChildProcess._handle.onexit (node:internal/child_process:286:19)
    at onErrorNT (node:internal/child_process:484:16)
    at process.processTicksAndRejections (node:internal/process/task_queues:82:21)] {
  errno: -2,
  code: 'ENOENT',
  syscall: 'spawn tslab',
  path: 'tslab',
  spawnargs: [
    'kernel',
    '--config-path',
    '/home/~/.local/share/jupyter/runtime/kernel-v2-29663rdSmjgt0v06u.json'
  ]
}
14:28:13.310 [error] 
14:28:13.310 [debug] Dispose Kernel process undefined.
14:28:13.313 [error] Failed to connect raw kernel session: Error: The kernel died. Error: ... View Jupyter [log](command:jupyter.viewOutput) for further details.
14:28:13.314 [error] Failed to connect raw kernel session: Error: The kernel died. Error: ... View Jupyter [log](command:jupyter.viewOutput) for further details.
14:28:13.314 [warn] Failed to shutdown kernel, .tslab./tslab./.tslab#kernel#--config-path#{connection_file} [TypeError: Cannot read properties of undefined (reading 'dispose')
    at R_.shutdown (/home/~/.vscode/extensions/ms-toolsai.jupyter-2024.6.0-linux-x64/dist/extension.node.js:304:13629)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
    at async A_.shutdown (/home/~/.vscode/extensions/ms-toolsai.jupyter-2024.6.0-linux-x64/dist/extension.node.js:304:22107)]
14:28:13.314 [trace] failed to create IJupyterKernelConnectionSession in kernel, UI Disabled = true [Ud [Error]: 
    at new n (/home/~/.vscode/extensions/ms-toolsai.jupyter-2024.6.0-linux-x64/dist/extension.node.js:98:4480)
    at new ml (/home/~/.vscode/extensions/ms-toolsai.jupyter-2024.6.0-linux-x64/dist/extension.node.js:181:13608)
    at new Ud (/home/~/.vscode/extensions/ms-toolsai.jupyter-2024.6.0-linux-x64/dist/extension.node.js:270:26489)
    at ky.launch (/home/~/.vscode/extensions/ms-toolsai.jupyter-2024.6.0-linux-x64/dist/extension.node.js:301:4)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)

Error: spawn tslab ENOENT
    at ChildProcess._handle.onexit (node:internal/child_process:286:19)
    at onErrorNT (node:internal/child_process:484:16)
    at process.processTicksAndRejections (node:internal/process/task_queues:82:21)] {
  category: 'unknown',
  isJupyterError: true,
  originalException: [Error: spawn tslab ENOENT
    at ChildProcess._handle.onexit (node:internal/child_process:286:19)
    at onErrorNT (node:internal/child_process:484:16)
    at process.processTicksAndRejections (node:internal/process/task_queues:82:21)] {
    errno: -2,
    code: 'ENOENT',
    syscall: 'spawn tslab',
    path: 'tslab',
    spawnargs: [
      'kernel',
      '--config-path',
      '/home/~/.local/share/jupyter/runtime/kernel-v2-29663rdSmjgt0v06u.json'
    ]
  },
  kernelConnectionMetadata: n {
    kind: 'startUsingLocalKernelSpec',
    kernelSpec: {
      specFile: '/home/~/.local/share/jupyter/kernels/tslab/kernel.json',
      name: 'tslab',
      argv: [Array],
      language: 'typescript',
      executable: 'tslab',
      display_name: 'TypeScript',
      metadata: [Object]
    },
    interpreter: undefined,
    id: '.tslab./tslab./.tslab#kernel#--config-path#{connection_file}'
  },
  stdErr: '\n\n'
}
14:28:13.316 [warn] Error occurred while trying to start the kernel, options.disableUI=true [Ud [Error]: 
    at new n (/home/~/.vscode/extensions/ms-toolsai.jupyter-2024.6.0-linux-x64/dist/extension.node.js:98:4480)
    at new ml (/home/~/.vscode/extensions/ms-toolsai.jupyter-2024.6.0-linux-x64/dist/extension.node.js:181:13608)
    at new Ud (/home/~/.vscode/extensions/ms-toolsai.jupyter-2024.6.0-linux-x64/dist/extension.node.js:270:26489)
    at ky.launch (/home/~/.vscode/extensions/ms-toolsai.jupyter-2024.6.0-linux-x64/dist/extension.node.js:301:4)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)

Error: spawn tslab ENOENT
    at ChildProcess._handle.onexit (node:internal/child_process:286:19)
    at onErrorNT (node:internal/child_process:484:16)
    at process.processTicksAndRejections (node:internal/process/task_queues:82:21)] {
  category: 'unknown',
  isJupyterError: true,
  originalException: [Error: spawn tslab ENOENT
    at ChildProcess._handle.onexit (node:internal/child_process:286:19)
    at onErrorNT (node:internal/child_process:484:16)
    at process.processTicksAndRejections (node:internal/process/task_queues:82:21)] {
    errno: -2,
    code: 'ENOENT',
    syscall: 'spawn tslab',
    path: 'tslab',
    spawnargs: [
      'kernel',
      '--config-path',
      '/home/~/.local/share/jupyter/runtime/kernel-v2-29663rdSmjgt0v06u.json'
    ]
  },
  kernelConnectionMetadata: n {
    kind: 'startUsingLocalKernelSpec',
    kernelSpec: {
      specFile: '/home/~/.local/share/jupyter/kernels/tslab/kernel.json',
      name: 'tslab',
      argv: [Array],
      language: 'typescript',
      executable: 'tslab',
      display_name: 'TypeScript',
      metadata: [Object]
    },
    interpreter: undefined,
    id: '.tslab./tslab./.tslab#kernel#--config-path#{connection_file}'
  },
  stdErr: '\n\n'
}
14:28:13.361 [debug] Disposed Kernel process undefined.
14:28:13.396 [trace] Conda file is conda
14:28:13.404 [trace] Web view is ready to receive widget messages

[0xjgv]

It looks like the kernel replaces your username egg for ~ in the path 🤔 Jupyter extension changes in VSCode @yunabe?

[egg-juxt]

Yeah. I'm not completely sure if it is an issue of finding the kernel executable, or the config file.

It seems to be unable to find the executable, because of the syscall that caused the Exception? (see originalException data above: syscall: 'spawn tslab')

It does work when using the Deno kernel though:

18:12:53.392 [info] Process Execution: ~/.deno/bin/deno jupyter --kernel --conn /home/~/.local/share/jupyter/runtime/kernel-v2-2665438FY804s8yRH8M.json
    > cwd: //home/~/src/temp/typeorm-xtdb/src
18:12:53.392 [debug] Kernel process 32993.

[egg-juxt]

Solved by adding the following to my ~/.profile:

# Enable nvm if available
if [ -f ~/.nvm/nvm.sh ]; then
  source ~/.nvm/nvm.sh
fi

tslab starts with #!/usr/bin/env node, and the task started by VS Code (non-interactive) wasn't finding node in the PATH.

I wonder if there'd be any VS Code settings, or something could be done in tslab in order to locate a proper node executable?