Re: Framework support for darwin

[Devang Patel <dpatel at apple dot com>]

On Tuesday, August 26, 2003, at 8:25 PM, Mike Stump wrote:

> Here is the latest version of the framework patch.  I'm of the 
> considered opinion, the target hooks are not appropriate for cpplib.  
> The below patch still contains the hook code, but, notice the 
> conditional that we were trying to avoid in register_include_chains 
> can't be gotten rid of.  One cannot initialize targetm, as then you 
> need the full back end for fix-header, and the full backend can't 
> work, as fix-header isn't a frontend...
>
> So, should I de-hookize cpplib things for fix-header, or leave it the 
> way it is.  It will work, but I don't find it, uhm, nicer.
>
> Also, one cannot hookize the TARGET_C_INIT stuff, as that is the magic 
> to allow the target hooks to work in the first place.  Again, the 
> disease seemed preferable to me.
>
> <frameworkg.diffs.txt>

! #define CPP_SPEC 
"%{static:%{!dynamic:-D__STATIC__}}%{!static:-D__DYNAMIC__} 
-D__APPLE_CC__=666"

Darwin GCC has code in driver to extract value from version_string and 
set APPLE_CC. You may want to use it.

-isysroot changes system root for system headers. How about system 
framework headers?

--
Devang