This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] fix PR39701 testsuite failures
> This is one of the areas where an expression is not one of the types of
> constant expressions listed in the C standard, but does not violate any of
> the constraints on constant expressions, so implementations are permitted
> to accept such expressions in initializers. It does fall outside my
> formal models of constant expressions, but if those are implemented in
> future for initializers a pedwarn (or warning) rather than an error would
Ok, so the patch (now committed) that enables
-fdelete-null-pointer-checks for -O0 and -O1 is the right way to fix
this, instead of tweaking the initializer.