On Tue, Nov 11, 2014 at 15:23:16 +0100, Martin Pitt wrote: > Control: tag -1 unreproducible > > Julien Cristau [2014-11-03 14:11 +0100]: > > your package failed to build on the ppc64el buildd: > > > ERROR:src/umockdev.c:6963:umockdev_script_runner_hamming: assertion > > > failed: (d1.length == d2.length) > > https://buildd.debian.org/status/fetch.php?pkg=umockdev&arch=ppc64el&ver=0.8.8-1&stamp=1411924863 > > This suggests either a race condition or a strange memory layout > problem. I built the package on pastel (the ppc64el porter box) and > was not able to reproduce this problem at all (I ran the tests some 20 > times). > > Would be interesting to see whether this is reproducible on the > buildd? Could the ppc64el buildd maintainers perhaps try a give-back? > If this happens consistently, then we need to work out the difference > between pastel and ppc64el1.debian.net (kernel?), if it succeeds now, > then either it's a race condition (can we try running the tests > several times on the buildd?), or some build dep changed underneath us. > Looks like 0.8.8-2 built successfully on ppc64el. It failed on armel this time though.
Cheers, Julien
signature.asc
Description: Digital signature
