2013/12/3 Mike Stump <mikest...@comcast.net>: > On Dec 2, 2013, at 5:02 AM, Chung-Ju Wu <jasonw...@gmail.com> wrote: >> Perhaps I should have used the following description, which seems much >> better: >> >> +/* { dg-skip-if "Variadic funcs have all args on stack. Normal funcs have >> args in registers." { nds32*-*-* } "*" "" } */ > > Reads nicely, thanks. Also, if I do a port, and this test case fails, and I > read that and those facts apply to my port, I can just effortlessly go that > direction. To me, this is the best use of this information. Secondary would > be if people wanted to do a target_supports, it would be more clear to the > untrained why it was done in the first place.
Thanks for the guidance. I learned how to make a better description for a test case. :-) And thanks for the approval, the revised patch is now applied as Rev. 205625: http://gcc.gnu.org/r205625 Best regards, jasonwucj