chromedriver breaks with pnpm's side-effects cache

[zkochan]

You can't do anything about it right now but I wanted to ask your opinion how to best fix this on pnpm's side.

The issue is that pnpm uses side-effects cache for caching builds. So pnpm will not run the post install script of a package twice on the same machine. This works great with most of the packages and makes installations faster. However, chromedriver's scripts have different side-effects when some environment variables are set. Hence, when pnpm caches the result of one build, it might be invalid to reuse the same cache for a different build of chromedriver as it might be executed with other environment variables. Here is an issue that was created as a result of this.

There are a couple of ways we may fix this. I have described the possible solutions in this issue. I think the best solution would be to support some package.json field that would tell pnpm which environment variables may influence the build of chromedriver. In that case, pnpm would be able to save the build cache under a key that would include the values of those env variables. Would you agree to ship such additional metadata for better DX of pnpm users?

[giggio]

I don't think adding a field to package.json to fix this specific issue would be a problem. How would that look like?

[zkochan]

It would probably be a list of environment variable names that pnpm should include in the cache key. For instance:

{
  "name": "chromedriver",
  "version": "103.0.0",
   "pnpm": {
     "includedEnvVarsInSideEffectsKey": ["detect_chromedriver_version"]
   }
}

So pnpm would use a different cache for chromedriver, when detect_chromedriver_version is set to true.

[giggio]

That's fine, please send a PR and I'll merge it.
Also, please send the link to pnpm's docs, so I can get more information about how pnpm will use this, and can better review the PR.

[zkochan]

ok, once we decide on a solution and it gets implemented, I'll send a PR. Thanks.