Junio C Hamano <gitster@xxxxxxxxx> writes:
> I haven't looked at the breakage in 1007 yet, though.
This turned out to be almost trivial. A fix relative to an earlier
version is that the call to test_oid helper needs to explicitly ask
for SHA-1 variant, as the command invocation outside a repository
uses SHA-1 (not due to falling back to a hardcoded default, but by
an explicit fallback in the "git hash-object" itself.
I'll send a v5 of the whole series sometime later (if I have time it
may happen today but otherwise tomorrow).
Thanks.
diff --git a/t/t1007-hash-object.sh b/t/t1007-hash-object.sh
index 64aea38486..d73a5cc237 100755
--- a/t/t1007-hash-object.sh
+++ b/t/t1007-hash-object.sh
@@ -260,4 +260,10 @@ test_expect_success '--literally with extra-long type' '
echo example | git hash-object -t $t --literally --stdin
'
+test_expect_success '--stdin outside of repository (uses SHA-1)' '
+ nongit git hash-object --stdin <hello >actual &&
+ echo "$(test_oid --hash=sha1 hello)" >expect &&
+ test_cmp expect actual
+'
+
test_done
diff --git a/builtin/hash-object.c b/builtin/hash-object.c
index 82ca6d2bfd..c767414a0c 100644
--- a/builtin/hash-object.c
+++ b/builtin/hash-object.c
@@ -123,6 +123,9 @@ int cmd_hash_object(int argc, const char **argv, const char *prefix)
else
prefix = setup_git_directory_gently(&nongit);
+ if (nongit && !the_hash_algo)
+ repo_set_hash_algo(the_repository, GIT_HASH_SHA1);
+
if (vpath && prefix) {
vpath_free = prefix_filename(prefix, vpath);
vpath = vpath_free;
