adding a regression test for 526471 to try to understand the problem.

1 files changed, 26 insertions, 0 deletions

diff --git a/testsuite/regression/526471-pdebuild-fail-scripts.sh b/testsuite/regression/526471-pdebuild-fail-scripts.sh
new file mode 100755
index 0000000..c0df1de
--- /dev/null
+++ b/testsuite/regression/526471-pdebuild-fail-scripts.sh
@@ -0,0 +1,26 @@
+#!/bin/bash
+# test pbuilder executes the failure script on failure
+# run like:
+#   rm -r work/; mkdir work/ ; ./526471-pdebuild-fail-scripts.sh
+
+set -e
+mkdir work/526471
+mkdir work/526471-hook
+echo 'echo failpkg-hook-executed' > work/526471-hook/C01_failhook
+chmod a+x work/526471-hook/C01_failhook
+echo 'echo failpkg-F-hook-executed' > work/526471-hook/F01_hook
+chmod a+x work/526471-hook/F01_hook
+echo 'echo failpkg-A-hook-executed' > work/526471-hook/A01_hook
+chmod a+x work/526471-hook/A01_hook
+HOOKDIR=$(readlink -f work/526471-hook)
+
+cd work/526471
+dpkg-source -x ../../../random-manual-test-material/failpkg/failpkg_0.1.dsc
+cd failpkg-0.1
+
+# the next command fails, but the error code is eaten by pipe; tee succeeds.
+pdebuild --use-pdebuild-internal -- --hookdir "${HOOKDIR}" | tee ../526471.log
+
+# I want to check for the output content. C01_failhook should have
+# been ran and output there.
+grep 'failpkg-hook-executed' 526471.log
\ No newline at end of file