Junio C Hamano [off-list ref] writes:
As batch_object_cb() makes a call to batch_object_write() with
obj_name set to NULL, I do not think this change is defensible,
though.
This indeed seems to break t5313 when queued on top of 'master'; it
tries to run "git cat-file --batch-all-objects --batch-check" and
hits the exact codepath where a missing object is sent to the error
codepath without obj_name set to anything.
I guess we now have an existing test that you can mimic that
exhibits "missing" error? I do not know offhand if this test
already qualifies as the test coverage Phillip wanted to make sure
exists.
..... >8 .......... >8 .......... >8 .......... >8 .......... >8 .....
expecting success of 5313.3 'pack/index object count mismatch':
do_pack $object &&
munge $pack 8 "\377\0\0\0" &&
clear_base &&
# We enumerate the objects from the completely-fine
# .idx, but notice later that the .pack is bogus
# and fail to show any data.
echo "$object missing" >expect &&
git cat-file --batch-all-objects --batch-check >actual &&
test_cmp expect actual &&
# ...and here fail to load the object (without segfaulting),
# but fallback to a good copy if available.
test_must_fail git cat-file blob $object &&
restore_base &&
git cat-file blob $object >actual &&
test_cmp file actual &&
# ...and make sure that index-pack --verify, which has its
# own reading routines, does not segfault.
test_must_fail git index-pack --verify $pack
4+0 records in
4+0 records out
4 bytes copied, 0.000119221 s, 33.6 kB/s
error: packfile .git/objects/pack/pack-67be769e2843d598c78218852612520795998892.pack claims to have 4278190080 objects while index indicates 1 objects
error: packfile .git/objects/pack/pack-67be769e2843d598c78218852612520795998892.pack claims to have 4278190080 objects while index indicates 1 objects
--- expect 2023-03-03 23:11:37.504011940 +0000
+++ actual 2023-03-03 23:11:37.508012250 +0000
@@ -1 +1 @@
-fff0a2476aa5c8e60a3ef21cfc66e0cc670920be missing
+(null) missing
not ok 3 - pack/index object count mismatch