LGTM.com - false positive

[ljharb]

Description of the false positive

It complains that https://github.com/es-shims/es5-shim/blob/master/es5-shim.js#L2024 has no effect. However, in JavaScript, concatenating with a string has always had a potential effect - it invokes the operand's toString() method, which might have side effects and might throw. In this case, it's explicitly being used inside a try/catch to trigger an exception when the operand is a Symbol.

https://lgtm.com/projects/g/es-shims/es5-shim/snapshot/19e189f143ae56896a49a02c4edb02505310a1e3/files/es5-shim.js?sort=name&dir=ASC&mode=heatmap#x229e38c0aa84b3d4:1

[asgerf]

Hi @ljharb,

Thanks for your report. We deliberately flag code whose only potential effect looks unintentional, and the cases where a coercion happens for its side effects alone are so rare that it's best to clarify the intent with a suppression comment. I can see you've already done so for a few other tools, but unfortunately LGTM does not understand ESLint or jscs suppression comments.

So if you're willing to add yet another a suppression comment to your collection, it would look like this:

'' + str; // jscs:ignore disallowImplicitTypeConversion lgtm[js/useless-expression]

Alternatively, the whole query can be disabled by checking in an lgtm.yml file with these contents:

queries:
  - exclude: js/useless-expression

[ljharb]

I'm not willing to add either, unfortunately; it might be wise for LGTM to add support for eslint suppression comments, since that's the defacto standard in the ecosystem.

In this specific case, however, since it's inside a try/catch, i don't think your heuristic should flag it.

[asgerf]

The other issue you opened (#2621) has a try/catch surrounding the statement, and I'm considering how best to fix the FP there. There isn't a try/catch that will obviously catch the exception here.

[ljharb]

Ah, that's true, I'm intentionally trying to trigger the exception so users see it.