Changing an error code is usually semver-major but I could the case for treating this as a bug fix. @nodejs/tsc Thoughts?

Personally, I'd like to see us reduce the specificity of error codes in general. I think ERR_INVALID_ARG_TYPE is fine and we can get rid of ERR_INVALID_CALLBACK entirely. But that would definitely be semver-major and outside the scope of this PR.

+1 to label this as semver-major, changing an error code is a potential breaking change.

+1 to get rid of ERR_INVALID_CALLBACK in favor of ERR_INVALID_ARG_TYPE at some point. validateCallback is not even checking if the parameter is actually a callback, if you pass a class it won't throw:

validateCallback(class{}); // doesn't throw

Changing an error code is usually semver-major but I could the case for treating this as a bug fix.

Bug fixes can be server-major too, so I'm adding the label back.

I would be more than happy to remove the ERR_INVALID_CALLBACK error entirely and replace it with ERR_INVALID_ARG_TYPE, but wouldn't that still cause error code changes? Since some APIs still seem to be using/throwing the ERR_INVALID_CALLBACK error such as the process.nextTick() method.

validateCallback is not even checking if the parameter is actually a callback, if you pass a class it won't throw:

validateCallback(class{}); // doesn't throw

That's the exact same case with the validateFunction() validator since both of the validateFunction() and validateCallback() validators have the exact same check, only difference being that they throw a different error.

We could entirely remove the validateCallback() validator and the ERR_INVALID_CALLBACK error and replace it with the validateFunction() validator, but we should change the behavior of the validator for it to be able to a differentiate a class and a function, so it would also throw on classes.

would be more than happy to remove the ERR_INVALID_CALLBACK error entirely and replace it with ERR_INVALID_ARG_TYPE, but wouldn't that still cause error code changes?

Yes, that would be a semver-major change (but one that is worth making in my opinion, especially if we can remove a few other codes in the same semver-major bump).

#41678