Hello,
Kurt Roeckx, on Sat 03 Mar 2012 22:10:11 +0100, wrote:
> On Sat, Mar 03, 2012 at 09:12:13PM +0100, Michael Biebl wrote:
> > The current version of elfutils FTBFS on several architectures:
> > hurd-i386
>
> Yes, nothing new. It should never have built on hurd, but it
> seems some people think it's a good idea to run buildds that
> don't run the testsuite.
We do not run buildds that don't run the testsuite.
We manually make some builds without tests when the failing tests are
limited and that unlocks a lot of other package builds. If we had to
always wait for testsuites we'd never get anywhere.
Kurt Roeckx, on Thu 19 Jul 2012 18:54:01 +0200, wrote:
> I'm also happy with a patch that does the right thin on hurd that
> doesn't use /proc/$pid/maps.
The maps file was added, it however for now lacks the paths.
Perhaps we can apply the attached patch to skip the only failing test,
as is done on kfreebsd?
Samuel
Index: elfutils-0.165/tests/run-native-test.sh
===================================================================
--- elfutils-0.165.orig/tests/run-native-test.sh
+++ elfutils-0.165/tests/run-native-test.sh
@@ -83,6 +83,9 @@ native_test()
# "cannot attach to process: Function not implemented".
[ "$(uname)" = "GNU/kFreeBSD" ] && exit 77
+# hurd's /proc/$PID/maps does not give paths yet.
+[ "$(uname)" = "GNU" ] && exit 77
+
native_test ${abs_builddir}/allregs
native_test ${abs_builddir}/funcretval
