https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83584
--- Comment #15 from Manuel López-Ibáñez <manu at gcc dot gnu.org> --- (In reply to Andrew Pinski from comment #11) > (In reply to Manuel López-Ibáñez from comment #8) > > I'll defer to Joseph or any other C FE maintainers to make the final call. > > https://gcc.gnu.org/ml/gcc-patches/2004-01/msg00599.html > > See that email about how this is invalid conversions from JSM already. The meaning of -Wpedantic has been refined over time and not all warnings under -Wpedantic are pedwarns (but they should be) and not all pedwarns are controlled by -Wpedantic (but they should be controlled by some flag). Also -pedantic-errors is not the same as -Werror=pedantic (but it should be a superset, which currently is not). The questions that decide whether some diagnostic is an error with -pedantic-errors are: * Does the standard says that such conversions should be rejected or require a diagnostic? My understanding is that the answer is No. * Does any potential undefined behaviour happen at compile-time or runtime? My understanding is that the answer is Compile-time. This seems the same case as -Wshift-negative-value: https://gcc.gnu.org/ml/gcc-patches/2015-04/msg01551.html Hence, the warning should not be an error with -pedantic-errors, hence, it should not be a pedwarn(), hence, it should not be a warning controlled by -Wpedantic. It can be still a warning, controlled by another flag, not enabled by -Wpedantic. (In reply to Keith Thompson from comment #12) > Is my understanding correct? The behaviour of the flags is detailed in the manual since a few years: https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html#index-pedantic-errors-1 Thus, -pedantic-errors will also give errors for compile-time UB and "in some other cases that do not prevent compilation of programs that are valid according to the standard"
