[odeint] Regression tests fail for unknown reason

On some machines all of odeints regression tests in the release branch fail for an unknown reason (http://www.boost.org/development/tests/release/developer/numeric-odeint.html). Does anyone know what happens here?

On 13 December 2012 20:03, Karsten Ahnert <karsten.ahnert@googlemail.com> wrote:
On some machines all of odeints regression tests in the release branch fail for an unknown reason (http://www.boost.org/development/tests/release/developer/numeric-odeint.html). Does anyone know what happens here?
It looks like it's an issue with Boost.Test.

Le 13/12/12 21:17, Daniel James a écrit :
On 13 December 2012 20:03, Karsten Ahnert <karsten.ahnert@googlemail.com> wrote:
On some machines all of odeints regression tests in the release branch fail for an unknown reason (http://www.boost.org/development/tests/release/developer/numeric-odeint.html). Does anyone know what happens here? It looks like it's an issue with Boost.Test.
Maybe, but I don't think so. All the Boost.Thread test are failling, and most of them don't use Boost.Test. The problem is that here is no information. What is happening with the regression test results? Best, Vicente

On Thu, Dec 13, 2012 at 9:21 PM, Vicente J. Botet Escriba <vicente.botet@wanadoo.fr> wrote:
Maybe, but I don't think so. All the Boost.Thread test are failling, and most of them don't use Boost.Test. The problem is that here is no information. What is happening with the regression test results?
Maybe backwards compatibility wouldn't have been so bad after all? -- Olaf

On 14 December 2012 10:45, Olaf van der Spek <ml@vdspek.org> wrote:
On Thu, Dec 13, 2012 at 9:21 PM, Vicente J. Botet Escriba <vicente.botet@wanadoo.fr> wrote:
Maybe, but I don't think so. All the Boost.Thread test are failling, and most of them don't use Boost.Test. The problem is that here is no information. What is happening with the regression test results?
Maybe backwards compatibility wouldn't have been so bad after all?
Not really, the problem was that the test results weren't displaying the error messages. If they had, then the fix would have been trivial.
participants (4)
-
Daniel James
-
Karsten Ahnert
-
Olaf van der Spek
-
Vicente J. Botet Escriba