On Mon, Sep 26, 2011 at 7:33 PM, Jason Merrill <ja...@redhat.com> wrote: > cxx_eval_logical_expression was assuming that a folded first operand of && > would be either boolean_true_node or boolean_false_node, but in fact it can > be a constant with a typedef of bool, which doesn't compare equal with ==. > So we should use tree_int_cst_equal to compare them instead. > > Tested x86_64-pc-linux-gnu, applying to trunk. > Thanks.
It is weird though that GCC does not maintain a properly typed internal representation. -- Gaby
