[PATCH] New pragma handling mechanism.

[Mark Mitchell]

Zack Weinberg wrote:

> Also, you need to implement the c_pragma_unblock_tokens() function
> that Richard Henderson wanted.  This function, if called when the
> current token is the CPP_EOF token that ends a pragma, consumes that
> token so the pragma handler can see subsequent file content.  It is an
> error to call it at any other time.

I'd really like to avoid the pragma handlers seeing subsequent file 
content.  I think that's a design mistake that will make life more 
difficult for us in future.  I don't think that it would be 
substantially harder to take a more structured approach now -- but if we 
don't do it now, it will inevitably be somewhat hard to switch later.

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304