Re: [RFC] Suggested replacement for specs and switch handling

[Neil Booth <neil at daikokuya dot demon dot co dot uk>]

Tom Tromey wrote:-

> gcj doesn't have a hard requirement that specs exist.  It just needs
> certain functionality.  If the same goals can be achieved in some
> other way, I'm fine with it.
> 
> The three problems I see are:
> 
> * Stuffing `-l' options into the link line
> * The possibility of running jvgenmain and cc1 at link time when
>   `--main' is given (presumably not hard as we already run other
>   programs conditionally)
> * The `combine inputs' code, which invokes a single jc1 instance on a
>   bunch of source files.
> 
> You'd know better than me whether these things are possible in the new
> scheme.

Well, we are going to need some front-end specific hooks in order to
do the equivalent of what, say, %(cc1_options) or CCP_SPEC do at
present.  Inside such a hook, I think all the above present no
problems at all :-) Good, I feel happier now.

Neil.