> Thanks again for the feedback.  Attached is a new version of the patch,
> with the complaints added.  Since I also added a complaint for the
> non-nested case, I also extended the testcase to cover that case as
> well.

Yet another version: Instead of generating the complaint while creating
the psymbols, I decided to generate them during the symtab creation
phase. A couple of reasons:
  - From an implementation perspective, it means that I only have to
    generate the complete from one location in our code.
  - Instead of generating a complaint for all broken DIEs all at once
    at startup, just generate them when it affects a unit which the user
    is actually referring to.
  - If we ever get rid of the psymtab, we'd lose the warning...

That's why I like this version better...

-- 
Joel