On Monday, 19. February 2007 16:15, Gabriel Dos Reis wrote:
> Please factor the hunks in c-typeck (parser_build_binary_op) and
> cp/call.c (build_new_op) into
> a separate function. maybe_warn_for_suspicious_logical_operator?
>
> Patch OK with those changes.
Hmm, you can't approve both patches at the same time ;) Manuel's patch
removed -Walways-true (which I didn't like anyway, so no objection at all
from me).
That leaves me behind with having to invent a new option for this warning. As
we now have -Waddress, how about adding -Woperator (where I would add this
warning) and -Wcomparison (where all the "always true due to range of
datatype" warnings would be)?