On Sat, 2015-06-20 at 13:37 -0700, Mike Stump wrote: > On Jun 19, 2015, at 5:36 PM, David Edelsohn <dje....@gmail.com> wrote: > > Maybe you should ask Richi or Jakub about the testcase because you are > > placing it in a non-target-specific location. It should succeed on > > all targets, but it may expose latent bugs on other targets. > > A latent bug is one that is broken, but appears to work, and then a change in > the compiler is made to expose the bug that was always there but didn’t show > a failure. It only applies to the compiler proper. > > In this case, since the change is to add a test case, the test case can only > show bugs that aren’t latent (or failures in the test case).
Yes, I felt that in this case we seem to have C++14 adding some new code paths that could well cause surprises on other targets besides Power. It seems like the sort of thing that language-specific torture tests are for, but please let me know if that's wrong and I'll move it to a target directory. Thanks, Bill
