For clarification, are you asking what it would take to to get file/line info for the resolution phase errors? Or are you asking how to make those kinds of errors occur?
If the former, we would need to evaluate how to store that information in a way that it would remain accessible through the whole parsing/resolution process. I can have some discussions about the feasibility and get back to you.
If the latter, anything that requires looking up another structure or field by identifier is going to be main source of resolution phase errors. Invalid symbols, incorrect types of symbols, etc.