Martin Michlmayr writes: > * Martin Buck <[EMAIL PROTECTED]> [2006-03-20 10:03]: > > I would have thought that a compiler that miscompiles trivial for loops and > > thus inserts bugs in lots of programs compiled with it is unsuitable for > > release, i.e. severity serious. > > I don't know... I compiled >5000 packages and none showed a problem. > Then again, most of them probably don't have a test suite as good as > your package (or a test suite at all); so they might compile and be > broken.
it appears that these loops are not that "trivial" ;-) Current practice is to consider these reports as non RC, if there's a workaround available, which currently is (use -fno-tree-loop-optimize), although it's arguable with wrong-code class bugs. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
