Re: [PATCH 1/2] unwritable tests: assert exact error output

Ævar Arnfjörð Bjarmason  [off-list ref] writes:

In preparation for fixing a regression where we started emitting some
of these error messages twice, let's assert what the output from "git
commit" and friends is now in the case of permission errors.

Usually we frown upon expecting an exact error message, but I guess
that the nature of the bug this series tries to address justifies
it.

-	test_must_fail git write-tree
+
+	cat >expect <<-\EOF &&
+	error: insufficient permission for adding an object to repository database .git/objects
+	fatal: git-write-tree: error building trees
+	EOF
+	test_must_fail git write-tree 2>actual &&
+	test_cmp expect actual
 '

OK.

 test_expect_success POSIXPERM,SANITY 'commit should notice unwritable repository' '
 	test_when_finished "chmod 775 .git/objects .git/objects/??" &&
 	chmod a-w .git/objects .git/objects/?? &&
-	test_must_fail git commit -m second
+
+	cat >expect <<-\EOF &&
+	error: insufficient permission for adding an object to repository database .git/objects
+	error: insufficient permission for adding an object to repository database .git/objects
+	error: Error building trees
+	EOF

This is odd.  Shouldn't the test expect one message from write-tree
and be marked as expecting a failure until the bug gets fixed?

+	test_must_fail git commit -m second 2>actual &&
+	test_cmp expect actual
 '