Re: GCJ on Darwin/i386 patches

[Sandro Tolaini <tolaini at libero dot it>]

On 28/mar/2006, at 20:07, Mike Stump wrote:

> One fix is to have a deployment target of `current' or some other  
> tag, and to change the default deployment target to `current', then  
> everything just works.  Current then selects the just installed  
> libgcc_s.dylib.  If people want java to work while targeting an  
> older release, they'll have to submit a patch to add the missing  
> symbol to a .a-style library and then the default default can be  
> set back to that release.  Geoff prefers the later bug fix.  I'd  
> prefer that if the design point is to be the later style fix, then  
> libgcc maintainers declare their intentions and state that any such  
> change is wrong and not permitted to libgcc.  To date, I don't  
> think that is the design point for libgcc.  And, if it isn't, then  
> darwin will continually be hit with these issues and have to `fix'  
> them one by one as they come up, which would just be annoying.   
> Hopefully, if I have misunderstood the libgcc design, someone will  
> fix libgcc and how it is used in the build.

I still haven't seen my patches committed to the trunk. I'd propose  
to leave the libgcc stuff aside for the moment, and commit all the  
other patches. In the meantime, we could figure out how to fix the  
missing symbol stuff.

Cheers,
  Sandro Tolaini

Attachment: smime.p7s
Description: S/MIME cryptographic signature