Re: flag_traditional relocation appropriate?

[Craig Burley <burley at gnu dot org>]

>  In message <199806160053.UAA21821@melange.gnu.org>you write:
>  > Should the flag_traditional global be moved from being defined in
>  > the C front end to be defined in toplev.c or some such place?  And
>  > the declaration accordingly moved into toplev.h?
>  > 
>  > Currently it seems the C front end defines and declares this
>  > almost exclusively *except* for the dwarf* modules, which are
>  > not part of the C front end.
>Hmmm, it really doesn't have a whole lot of meaning outside of
>the C language.  What would "traditional" C++ or Fortran mean? :-)

Well, it *can* have a useful (and consistent) meaning for Fortran at
least, but that's not the point.

The point is, since the option doesn't necessarily exist for a given
front end, why should every front end have to define the global in
its source code?

>It looks like the dwarf* modules use it to determine what C variant
>to record in the dwarf debug record since apparently dwarf and dwarf2
>distinguish between "old C" and c89.

There's certainly a kludge (at least one ;-) in that.  And maybe
this kludge isn't easily dealt with (for 1.1, for example).

But with all the cleaning up of the gcc header files going on, I
noticed this wart and thought I'd mention it.  Wouldn't surprise
me if someone looked at it and figured out a way either to make
it entirely back-end-based or to remove it to only the C front end.

        tq vm, (burley)