------- Comment #10 from pinskia at gcc dot gnu dot org 2006-10-27 01:28 ------- (In reply to comment #9) > one of those patches introduces the regression on the 4.1 branch
The backport of Nathan's patch caused it, though DR 224 is the questionable problem here and reading the standard currently (without that Defect report), this is actually invalid code. The testcase in PR 29469 comment #5, shows that Defect report is bogus and needs to be rethought. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29607
