Hi,
On Thu, 10 Jul 2025, at 14:56, Paul Gevers wrote:
> Your package has an autopkgtest, great. However, it fails since January
> 2025. Can you please investigate the situation and fix it? I copied some
> of the output at the bottom of this report.
Thanks for reporting this. I understand this has happened due to some
combination of unrelated changes around GNU make and debhelper. Bmake ends up
being called with the following in the environment:
MAKEFLAGS=-w -j10 --jobserver-auth=fifo:/tmp/GMfifo2724
This is, obviously, not going to work, as bmake doesn’t support *any* long
(dash-dash) options, and does not support job servers at all. Which means the
debhelper integration module has to filter out all long options out of
MAKEFLAGS:
@@ -56,6 +56,10 @@ sub do_make {
if ($ENV{MAKEFLAGS} =~ /^[^-]/) {
$ENV{MAKEFLAGS} = "-$ENV{MAKEFLAGS}";
}
+ my @makeflags = grep {
+ !/^--/
+ } split(' ', $ENV{MAKEFLAGS});
+ $ENV{MAKEFLAGS} = join(' ', @makeflags);
}
> With my Release Team member hat on I allow you to tag this bug as
> trixie-ignore if this is nearly completely a test issue as it's so late
> in the freeze and it's not worth removing the package from trixie
> because of a failing test at this moment unless it exposes a much bigger
> issue. Having said that, if this is a test only issue and a fix is
> possible without fully removing the autopkgtest and without making the
> test superficial, it's still welcome, but it would need to happen soon.
I would prefer to fix the bug in the debhelper integration, since leaving it
unfixed may also break other packages relying on bmake, and we may have just
not triggered those errors yet.
Should I submit an unblock request? The changes will be just the above diff
plus the changelog entry.
--
Cheers,
Andrej
