[QUESTION] Alternative for $npm_package_* to pass multiple parameters to node

[dandv]

Motivation ("The Why")

In my package.json, I have several scripts that call node with the same parameters over and over. To simplify that invocation, I've created a package.json variable, and am invoking the scripts like this:

{
  "nodeParams": "--experimental-specifier-resolution=node --harmony -r source-map-support/register",
  "scripts": {
    "dostuff": "node $npm_package_nodeParams myscript.js"
  }
}

This is quite common for calling compiled TypeScript code, but no longer works in v7 due to RFC21. However, I can't find what the recommended alternative is.

I picked up this idea from somewhere on StackOverflow, so I'm probably not the only one impacted by the change.

[ivpcode]

hi, same issue.

[artischockee]

Same issue. I used to retrieve data from author.name field (in $npm_package_author_name env variable) and it was being revealed in my app. But, perhaps starting from this "RFC21", this variable is not showing at all, giving the consumer an empty string.

[isaacs]

You can put these values in a config block in package.json, like this:

{
  "config": {
    "nodeParams": "--experimental-specifier-resolution=node --harmony -r source-map-support/register",
    "author": {
      "name": "Wile E. Coyote"
    }
  },
  "scripts": {
    "dostuff": "node $npm_package_config_nodeParams myscript.js",
    "whoami": "echo $npm_package_config_author_name"
  }
}

Passing the entirety of package.json into the environment was contributing to problems, because sometimes these files contain a lot of stuff, including very long strings that blew up the environment size. (We were also passing every last config value, even those set at the default value, so that was also adding to the problem.)

[wu-lee]

Thanks - this mechanism also provided access to parameters from the .npmrc file(s). What is the recommended forward and backward compatible way of getting access to those again?

Also, this information is not easy to find. See npm/cli#2452 (comment)