Driver rewrite... (was Re: RFC: should we use -Werror?)

[Zack Weinberg]

On Wed, Sep 05, 2001 at 07:12:08PM +0100, Neil Booth wrote:
> Zack Weinberg wrote:-
> 
> > I think that if you show up with a
> > compelling proof-of -concept implementation that does 80% of what
> > people want, faster and neater, they'll be willing to work out
> > better ways of doing the other 20%.
> 
> It would be a lot of work to get it to that stage.  For example, to
> get something suitable for a commit would mean touching every port,
> both for coding up routines to handle port-specific flags and to
> remove vairous stuff like all the macro gunk that currently exists.
> There's a lot of it, and it's repetetive, dull and can't be automated.

Um, what was wrong with my suggestion that you put the new driver in a
separate file and add Makefile+config.gcc bits so that each port can
choose which one it wants?  Then ports can be converted incrementally.
You have to touch all the front ends, but there are only five of them
in the tree.  (Which reminds me, what happened to Ada?  It was going
to be put into the tree in July, but then it wasn't.)

As for the runtime configurability, I very much think we were all
waving hands frantically at each other because your proposal was vague
enough that it was hard to tell what it could do.  Come back with
concrete code and at least we can argue about something solid...

zw