https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108555
--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> --- (In reply to Iain Sandoe from comment #2) > We do claim the ones that modula-2 actually uses in gm2-lang.opt > > (previous to r13-5373-g80cf2c5e8f496b we had a large and growing set to > accommodate everything that the preprocessor needs). > > Because of that growing number of added options, this change was made > intentionally... the idea was to claim and act on the C and Driver ones that > are used to construct C-preprocessor command lines as an action separate > from handling the Modula-2 options. > > IIUC, the mechanism of adding them to the language opt is a convention > rather than a correctness issue? Kind-of ... > Or is it that we should reject rather than ignoring the options that we do > not pass to the preprocessor? ... yes, you are short-cutting generic code that diagnoses and removes unappropriate options, I'm not sure simulating that behavior is easily possible. > (I can revert to adding everything to the gm2-lang.opt .. it just seemed a > bit hacky)
