On 12-01-18 01:11 PM, Eric Anholt wrote: > On Tue, 17 Jan 2012 15:58:11 -0500, Matt Turner <matts...@gmail.com> wrote: > > As far as I can tell, the point of AM_CPPFLAGS existing is if you want > to make a target that separately calls the preprocessor, so you a list > of includes and a list of non-preprocessor compiler flags. Given that > we don't have any preprocessor invocations, we're sure to leak actual > cflags into the cppflags, so I don't see why we'd make this change.. That's the only scenario I am aware of that will actually "break". You never know when this target might be added in the makefile. It's preferable to keep the flags separate where possible. Automake keeps them separate from top to bottom. A builder can append flags at make invocation using CFLAGS and CPPFLAGS, knowing they will used with the correct target.
Unfortunately too many developers don't bother and when mixed flags are passed in through a variable from another module, they cannot be separated again. In another post, I expressed my reservations due to INTEL_CFLAGS that may contain mixed flags. Using AM_CFLAGS would be safer, unfortunately. _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
