On Wed, 13 Dec 2006, Paolo Bonzini wrote: > > > I'd still like to drop the ".tab". As long as we say %language is > > experimental, we can change our minds later if it causes trouble. > > I think this would complicate a bit the patch, because I wouldn't be able to > use valid_languages[0] as the default.
I think you could. Just ignore add_tab if language_prio < 2. But.... > I'd rather leave this for later, > possibly for when we implement the language.m4 idea That's fine. > > What if we merge the valid_languages table (appearing in src/getargs.c) with > > data/c++-skel.m4 and data/c-skel.m4 to produce a single data/language.m4? > > This kind of single config script for %language would make it easier for > > users and developers to add support for new languages and skeletons without > > rebuilding Bison. I think it would also simplify your patch and reduce > > %language's footprint in the Bison front-end source. > > This I'm not sure, but the idea is worth giving a whirl... Great. > > What do you think? > > ... however, I'd rather do this later, because I see this as a cleanup. I was hoping that it would simplify your patch significantly, and then Paul Eggert might be more inclined to review it before you commit it. I find it hard to understand a series of successively applied patches where a later one significantly rewrites the logic of an earlier one. > I > also need some time to digest your latest @output patch (which would ease the > implementation too) Yep. > and to see how this idea would interact with %skeleton, > then I'll come back. That's fine.
