%Z only uses seconds, so it's possible that we did the commit
in the same second, which made this test racy.
- Switch to full nanosecond precision using '%.Y' so it always differs
- Fix the inverted `cmp` usage
- Add a missing `ok`
set -euo pipefail
-echo "1..49"
+echo "1..50"
$OSTREE checkout test2 checkout-test2
echo "ok checkout"
mkdir -p test2-checkout
cd test2-checkout
touch blah
-stat --printf="%Z\n" ${test_tmpdir}/repo > ${test_tmpdir}/timestamp-orig.txt
+stat --printf="%.Y\n" ${test_tmpdir}/repo > ${test_tmpdir}/timestamp-orig.txt
$OSTREE commit -b test2 -s "Should bump the mtime"
-stat --printf="%Z\n" ${test_tmpdir}/repo > ${test_tmpdir}/timestamp-new.txt
+stat --printf="%.Y\n" ${test_tmpdir}/repo > ${test_tmpdir}/timestamp-new.txt
cd ..
-if ! cmp timestamp-{orig,new}.txt; then
+if cmp timestamp-{orig,new}.txt; then
assert_not_reached "failed to update mtime on repo"
fi
+
+echo "ok mtime updated"