On Tue, Mar 1, 2011 at 11:33 AM, Tom Rini <[email protected]> wrote: > On 03/01/2011 12:14 PM, Khem Raj wrote: >> >> On Tue, Mar 1, 2011 at 10:29 AM, Tom Rini<[email protected]> wrote: >>> >>> On 03/01/2011 11:10 AM, Phil Blundell wrote: >>>> >>>> On Tue, 2011-03-01 at 10:54 -0700, Tom Rini wrote: >>>>> >>>>> That's why I'm asking for the -g part to be put into a variable so that >>>>> it can be overriden easily, in all cases, without needing to know the >>>>> special cases (ie glibc). >>>> >>>> Why is glibc special anyway (or, why is the optimization in a >>>> libc-specific conf file)? It seems a bit bogus that the choice of your >>>> C library would be having such a global impact on the compiler flags for >>>> everything else. >>> >>> There is some odd papering over that's going on to switch out a flag or >>> two >>> (and force -O2) for glibc (but not eglibc). But my point is that today >>> we >>> have a few special case FULL_OPTIMIZATION_pn-'s and it'd be great to just >>> not have to worry about that when setting the debug level we build with. >>> >> >> May be debugability can be expressed via another variable say >> DEBUG_FLAGS or somesuch which can then we appended to global CFLAGS > > With the caveat that I want to control native vs target at different levels, > yes.
we can have a native counter part or calculate it using native override > > -- > Tom Rini > Mentor Graphics Corporation > _______________________________________________ Openembedded-devel mailing list [email protected] http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
