Dear all, As one of the developers of ProbABEL I can confirm this is a rounding error. We have had similar problems in the past. The test run by make is simply diff-ing the output of a test run with a 'known good' file. Back when ProbABEL was still actively developed, I tried my best to maintain accuracy across the platforms I knew were being actively used by our users. ARM and PPC where never among those, neither did I have access to machines with that architecture.
All in all I think either of Andreas' proposed solutions is fine. Best regards, Lennart. On 02-02-19 12:19, Andreas Tille wrote: > On Sat, Feb 02, 2019 at 11:56:50AM +0100, Thierry fa...@linux.ibm.com wrote: >> Hi, >> >> Thanks for pointing out that my analysis wasn't good - the arm patch >> doesn't change anything for ppc64el (which make sense whatsoever) - the >> problem on ppc64el is similar to the one on arm except the test value >> are opposite ! ( which was confusing). >> So what do we need to do ? expect that someone with a better expertise >> help going though it ? ( can you confirm we stay with that bug for both >> arch or do we need to open one for ppc64el ?) > > As long as we are sure that its just a rounding error I think an apropriate > solution would be to ignore the error for ppc64el (either exclude the test > or run > > dh_auto_test || true > > to be able to inspect the log. > > Do you think that's in your interest? > > Kind regards > > Andreas. > -- *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* L.C. Karssen 's-Hertogenbosch The Netherlands lenn...@karssen.org http://blog.karssen.org GPG key ID: A88F554A -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
signature.asc
Description: OpenPGP digital signature