This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Fix warnings and errors building libiberty with pcc on vax-dec-ultrix4.3
> The vax CPP behaves the same. So, the md5.c code is ok and the gcc
> warning is bogus because it comes in the ISO preprocesor which
> pre-expands arguments.
>
> On a slightly different subject, I noticed in reworking the PA
> backend to use TARGET_OS_CPP_BUILTINS and TARGET_CPU_CPP_BUILTINS,
> that builtin_define_std still adds a traditional predefine when
> -ansi is not specified.
>
> The HP C compiler on hpux has three levels: classic, ansi-extended
> and ansi. The PA backend has traditionally provided a namespace
> similar to the HP class namespace except that __CLASSIC_C__ is
> not defined and __STDC_EXT__ is defined. GCC also adds some
> extra ansified symbols that aren't in the namespace of the HP
> compiler. Thus, in effect, we have a standard C compiler using
> a traditional namespace.
>
> I am tempted to change the namespace under hpux 10.X and 11 to
> make it closer to the ansi-extended namespace. This would entail
> not defining hppa, unix and various PWB symbols.
>
> I tried just defining the traditional symbols when the option
> -traditional-cpp was given but the struct cpp_reader is incomplete
> in c-common.c where the above TARGET macros are expanded.
>
> Any thoughts on what to do about traditional predefines?
You've lost me completely I'm afraid. What's a "traditional predefine"?
Neil.