Mattia Rizzolo writes ("Bug#903441: dgit: autopkgtest failures in Ubuntu"):
> I'll wait for a debdiff to apply on top of 5.8 :)
I am confused by what I think is the behaviour of dch in this test.
In the test, dgit tries to upload to the suite bionic in the distro
test-dummy. The suite is supposed to come from the changelog. The
dch invocations were:
+ faketime @1515000000 dch -v 2.0-1 -m 'new upstream (did gbp import-orig)'
+ faketime @1515000000 dch -r sid
The first one of these ought to make an UNRELEASED entry AFAICT (even
reading bionic's dch). The second an entry targeting sid.
I tried to repro this by copying debchange.pl from bionic's devscripts
into my /usr/bin and running the test with DEBCHANGE_VENDOR=Ubuntu.
But it passed.
Can you run the test on bionic with this attached patch ?
Note that this will cause all of the working stuff from the tests to
turn up in AUTOPKGTEST_ARTIFACTS. If you run this in a fully formal
way on the whole test suite, the result will be about 300Mby. If that
is too much, run just --testname=gdr-makepatches7.
Alternatively, you can run this in an ad-hoc way out of the build
tree. (i) dgit clone dgit (ii) apply patch (iii) install the
test-dependencies for the gdr-makepatches7 test (which are all
harmless to have on your system)
(iv)
tests/using-intree tests/tests/gdr-makepatches7 2>&1 |tee log
Ian.
diff --git a/tests/lib b/tests/lib
index 4ef275ca..6ee6133d 100644
--- a/tests/lib
+++ b/tests/lib
@@ -40,6 +40,10 @@ testname="${DGIT_TEST_TESTNAME-${0##*/}}"
export DGIT_TEST_TROOT=$troot
tmp=$ADTTMP
+if [ x"$AUTOPKGTEST_ARTIFACTS" != x ]; then
+ tmp=$AUTOPKGTEST_ARTIFACTS/tmp
+ mkdir -p -- "$tmp"
+fi
if [ x"$tmp" = x ]; then
export DGIT_TEST_BYHAND=1
mkdir -p tests/tmp
--
Ian Jackson <ijack...@chiark.greenend.org.uk> These opinions are my own.
If I emailed you from an address @fyvzl.net or @evade.org.uk, that is
a private address which bypasses my fierce spamfilter.
