It appears that the returned error is unused. And it is in fact unused, but `newError` also logs it, and panics if panicking on error is enabled. So, I think it works correctly and as intended and is just not super clear from the context.
This is probably the second instance where we use newError this way. Maybe we should add a reportError or logError function that calls newError under the hood but makes the intended behavior clearer?
Originally posted by @nickkhyl in #89 (comment)