> attached patch fixes PR 13092, so that -Wuninitialized no longer warns. > Does it look reasonable? LGTM. (did we not already have an equivalent test for the suppressed case (the first of your two test cases)?)