[FEATURE] Allow satisfies to throw errors

[jharrilim]

What / Why

It would be nice to have satisfies throw an error so that we may see that the range does not satisfy due to an invalid range as opposed to an invalid version. The range error that we'd like to see is here: https://github.com/npm/node-semver/blob/cb1ca1d5480a6c07c12ac31ba5f2071ed530c4ed/classes/range.js#L41

When

• When verifying engine versions during npm install
• When adding a dependency to package.json during npm install

Where

• npm/node-semver

How

Current Behavior

Create an empty project with the following package.json:

{
  "name": "foo",
  "version": "1.0.0",
  "description": "",
  "main": "index.js",
  "scripts": {
    "test": "echo \"Error: no test specified\" && exit 1"
  },
  "author": "",
  "license": "ISC",
  "engines": {
    "node": ">=6.5.0 !15"
  },
  "engineStrict": true
}

Run npm install and you get:

npm WARN EBADENGINE Unsupported engine {
npm WARN EBADENGINE   package: '[email protected]',
npm WARN EBADENGINE   required: { node: '>=6.5.0 !15' },
npm WARN EBADENGINE   current: { node: 'v16.11.1', npm: '8.0.0' }
npm WARN EBADENGINE }

up to date, audited 1 package in 414ms

found 0 vulnerabilities

Expected Behavior

Create an empty project with the following package.json:

{
  "name": "foo",
  "version": "1.0.0",
  "description": "",
  "main": "index.js",
  "scripts": {
    "test": "echo \"Error: no test specified\" && exit 1"
  },
  "author": "",
  "license": "ISC",
  "engines": {
    "node": ">=6.5.0 !15"
  },
  "engineStrict": true
}

Run npm install and ideally get something like:

npm WARN EBADENGINE Invalid SemVer Range: >=6.5.0 !15

up to date, audited 1 package in 414ms

found 0 vulnerabilities

Who

• n/a

References

• n/a

[lukekarrys]

I'm not opposed to this but I do think this would be a breaking change.

[AlanSl]

This would also be very useful for debugging when a bug or failure in semver is encountered, such as https://github.com/npm/node-semver/issues/381 or https://github.com/npm/node-semver/issues/354

Currently, many errors or failures within semver return a false negative with no clue that something has failed - unless a debugger with "break on caught errors" is active, they appear exactly like a completed validation that has successfully judged the version to be invalid.

If what actually happened was semver hit an error mid-validation due to some bundling issue or internal bug, it's important to be able to see that and treat it differently, rather than have an operation continue based on possibly incorrect information.

(we've just had exactly this happen, and it took a long time to figure out that there was a caught internal error in semver caused by a bundling issue)