Custom jupyter labextension fails to install - node.js

Question

We have a custom jupyterlab extension that fails to install for nodejs=14.14.0 (and for 15.2.1) (installed using conda).
Here's how the package.json file looks like for the labextension:
{
"name": "my-custom-ext",
"version": "0.1.0",
"description": "Integrate JupyterLab with ext",
"keywords": [
"jupyter",
"jupyterlab",
"jupyterlab-extension"
],
"homepage": "https://github.com/my_name/myextension",
"bugs": {
"url": "https://github.com/my_name/myextension/issues"
},
"license": "BSD-3-Clause",
"author": "xyz",
"files": [
"lib/**/*.{d.ts,eot,gif,html,jpg,js,js.map,json,png,svg,woff2,ttf}",
"style/**/*.{css,eot,gif,html,jpg,json,png,svg,woff2,ttf}"
],
"main": "lib/index.js",
"types": "lib/index.d.ts",
"repository": {
"type": "git",
"url": "https://github.com/my_name/myextension.git"
},
"scripts": {
"build": "tsc",
"clean": "rimraf lib",
"prepare": "npm run clean && npm run build",
"watch": "tsc -w"
},
"dependencies": {
"#jupyterlab/application": "^3.0.6",
"#jupyterlab/apputils": "^3.0.5",
"#jupyterlab/coreutils": "^5.0.3",
"#jupyterlab/docregistry": "^3.0.6",
"#jupyterlab/launcher": "^3.0.5",
"#jupyterlab/mainmenu": "^3.0.5",
"#jupyterlab/notebook": "^3.0.6",
"#jupyterlab/services": "^6.0.5",
"#lumino/commands": "^1.12.0",
"#lumino/coreutils": "^1.5.3",
"#lumino/disposable": "^1.4.3",
"#lumino/widgets": "^1.16.1"
},
"devDependencies": {
"#jupyterlab/builder": "^3.0.0",
"rimraf": "^3.0.2",
"typescript": "~4.1.3"
},
"jupyterlab": {
"extension": true
}
}
Here's the error jupyter labextension install --no-build my-custom-ext :
Node v14.14.0
Yarn configuration loaded.
> node /opt/conda/lib/python3.8/site-packages/jupyterlab/staging/yarn.js install
-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-yarn install v1.21.1
info No lockfile found.
[1/4] Resolving packages...
[2/4] Fetching packages...
warning #blueprintjs/core#3.52.0: Invalid bin entry for "upgrade-blueprint-2.0.0-rename" (in "#blueprintjs/core").
warning #blueprintjs/core#3.52.0: Invalid bin entry for "upgrade-blueprint-3.0.0-rename" (in "#blueprintjs/core").
error #npmcli/fs#1.1.0: The engine "node" is incompatible with this module. Expected version "^12.13.0 || ^14.15.0 || >=16". Got "14.14.0"
error Found incompatible module.
info Visit https://yarnpkg.com/en/docs/cli/install for documentation about this command.
> /opt/conda/bin/npm pack /home/jovyan/my-custom-ext
-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-\|/-
> my-custom-ext#0.1.0 prepare /home/jovyan/my-custom-ext
> npm run clean && npm run build
> my-custom-ext#0.1.0 clean /home/jovyan/my-custom-ext
> rimraf lib
> my-custom-ext#0.1.0 build /home/jovyan/my-custom-ext
> tsc
src/index.ts(1,72): error TS2307: Cannot find module '#jupyterlab/apputils' or its corresponding type declarations.
....
....
....
ValueError: "/home/jovyan/my-custom-ext" is not a valid npm package
Exiting application: lab
Running "npm ls" I got :
├─┬ #jupyterlab/builder#3.2.5
│ ├─┬ terser-webpack-plugin#4.2.3
│ │ ├─┬ cacache#15.3.0
│ │ │ ├─┬ #npmcli/fs#1.1.0
changed package.json to :
"devDependencies": {
"#jupyterlab/builder": "=3.2.4",......
But that didnt help. Still got the same error
Note : https://www.npmjs.com/package/#npmcli/fs :1.1.0 was published 5 days ago.
Please help how to fix

Answer 1

This fragment is relevant:
Expected version "^12.13.0 || ^14.15.0 || >=16". Got "14.14.0"
It tells you that you need to use node 14.15.0 or newer 14.x and that you cannot use 13.x, 15.x. But you have 14.14.0 - this means that you need to upgrade your node.js.
You mention conda - it might be that you use outdated default channels and therefore you get the old minor release of 14.x; conda-forge channel has all the versions and I recommend switching to it. If you already use conda-forge and cannot get the newer version, you may want to try using mamba as it has a better dependency resolver.
Note: 15.x and newer is not well supported by JupyterLab 3.x at the moment anyways (among other because not all dependencies upgraded to 16.x yet as Node follow odd-even version cadence where only even numbers are stable long enough for the wider ecosystem to have incentive to migrate).