Suggestions for how to run protoize over a huge source tree?

[Miles Lane]

Hi,

There are ~14,500 "function declaration isn't 
a prototype" warnings in the XFree86 source code.  
I would like to make a huge dent in moving the 
code base to ANSI function declarations.

Since there are a lot of different CCOPTIONS and 
include directories that vary depending on the 
subdirectory that is being compiled, I'd like to 
modify my makefiles so that instead of simply running 
gcc over each source file, protoize gets run first 
and then gcc makes a pass.  Ideally, there'd be a 
way to ignore all sorts of errors, so that at least 
one full protoize pass gets made in each directory.

I recognize that this may be a horrible plan.
If it is, please explain why and offer alternatives.

Since this is a huge codebase, hopefully this process
can be streamlined as much as possible.

Thanks,
	Miles