@@ -362,6 +362,28 @@ test_expect_success 'tree entry with duplicate type mismatching objects' '
)
'
+test_expect_success 'diff-tree stressing tree-diff.c::tree_entry_pathcmp(), not the same type' '
+ zero=$(test_oid zero) &&
+ git -C duplicate-entry diff-tree broken-commit-1 broken-commit-2 >1-to-2 &&
+ grep "$zero" 1-to-2 >lines &&
+ test_line_count = 2 lines &&
+
+ git -C duplicate-entry diff-tree broken-commit-2 broken-commit-1 >2-to-1 &&
+ grep "$zero" 2-to-1 >lines &&
+ test_line_count = 2 lines
+'
+
+test_expect_success 'diff-tree stressing tree-diff.c::tree_entry_pathcmp(), types not reversed' '
+ blob_ok=$(git -C duplicate-entry rev-parse broken-commit-2:A) &&
+ git -C duplicate-entry diff-tree --diff-filter=A broken-commit-1 broken-commit-2 >1-to-2 &&
+ grep "$blob_ok" 1-to-2 &&
+ test_line_count = 1 1-to-2 &&
+
+ git -C duplicate-entry diff-tree --diff-filter=A broken-commit-2 broken-commit-1 >2-to-1 &&
+ grep "$blob_ok" 2-to-1 &&
+ test_line_count = 1 2-to-1
+'
+
test_expect_success 'tag pointing to nonexistent' '
badoid=$(test_oid deadbeef) &&
cat >invalid-tag <<-EOF &&
Piggy-back on the recently added fsck tests for mode comparisons in mktree and assert that diff-tree also does the right thing in this implausible scenario. As with the other tests I've added in preceding commits, these tests will fail if the mode is the same or reversed, respectively. The diff-tree code being tested here was originally added back in .9174026cfe (Add "diff-tree" program to show which files have changed between two trees., 2005-04-09). Unlike the other tests I've added there are existing tests for both of these scenarios. Breaking that function as described above will make tests in t4002-diff-basic.sh, t6409-merge-subtree.sh and t4037-diff-r-t-dirs.sh fail. I think it's good to have tests for this regardless, so let's add these. Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> --- t/t1450-fsck.sh | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+)