[Debian GNUstep maintainers] Bug#449163: gnustep-base: FTBFS on arm: checking if ffcall trampolines work... no

Yavor Doganov yavor at gnu.org
Mon Nov 19 15:30:53 UTC 2007


On Wed, Nov 14, 2007 at 10:42:05AM -0500, Hubert Chathi wrote:
> 
> Based on this, it's probably best to just stick with ffcall for now, 
> until we really need to switch to libffi (if we ever need to).

Sounds reasonable, but how do you plan to resolve the current issue?  I 
see two ways:

1) Disable the arm patch in the ffcall Debian package and upload ffcall 
   + gnustep-base accordingly (ideally, this should be tried on an arm 
   development box...)

   1a) Report the bug to ffcall upstream.  IMVHO it would be useless to 
   do this now since that arm-specific patch is not present upstream.  
   It might take a long time to get ffcall fixed, though.

2) Build gnustep-base with libffi conditionally on arm.

3) Not really an option -- hack gnustep-base's build system to ignore 
   the test on arm.  This is potentially dangerous, as probably GNUstep 
   on arm has always been broken, so it might be a good time to fix this 
   now.  Hiding the bug is always wrong.

Of course, there's no guarantee that -base will build with ffcall with 
the disabled patch or with libffi.

-- 
I had a very low opinion of TCL, basically because it wasn't Lisp. --RMS





More information about the pkg-GNUstep-maintainers mailing list