------- Comment #8 from ian at airs dot com 2007-04-28 05:26 ------- I don't see why this PR should count as a regression, since there is no regression in the generated code. There is only a change in VRP behaviour. The generated code is the same. This PR is only going to be really interesting if we find a code regression.
-- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31130
