On 03/13/09 13:03, Laszlo (Laca) Peter wrote: > On Fri, 2009-03-13 at 10:46 +0000, Ghee Teo wrote: > >> If this patch is sparc only, apply that patch only when it is building >> for sparc, that can be easily handled in spec file, use >> %ifarch sparc >> > > However, it's ugly and makes it impossible to upstream the patch, > so please fix the patch instead so that it can be used on > sparc and x86. > Good point. I think Joerg has come up with a proposed fix in this thread :)
-Ghee > Laca > > >> Then the changes would not affect x86. Sinc eyou have not include the >> spec file changes in the review. So Joerg is correct in his query. >> >> -Ghee >> >> >> On 03/13/09 09:57, Jerry Tan wrote: >> >>> On 03/13/09 17:53, Joerg Schilling wrote: >>> >>>> Jerry Tan<Jerry.Tan at sun.com> wrote: >>>> >>>> >>>> >>>>>> If I did not miss something, you completely removed the >>>>>> --output-endianess=little option, this will cause it to fail on Intel >>>>>> hardware. >>>>>> >>>>>> Cdda2wav by default always outputs data in network byte order. >>>>>> >>>>>> If you request raw audio data, it is a good idea to write interfaces the >>>>>> same >>>>>> way and convert byte order lately before it is send to the driver (in >>>>>> case the >>>>>> driver is not able to do this abstraction from machine byte order). >>>>>> >>>>>> J?rg >>>>>> >>>>>> >>>>>> >>>>>> >>>>> Yes, that's what the patch try to do. >>>>> remove "--output-endianess=..." >>>>> >>>>> >>>> Then the patch breaks things on x86. >>>> >>>> J?rg >>>> >>>> >>>> >>> so I have to add #ifdef __sparc here? >>> >>> > >
