git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH] t/t1467-refs-rename: Use test_must_fail in place of NOT operator

From
Junio C Hamano <gitster@pobox.com>
Date
Oct 28, 2026, 03:43 UTC
Message-ID
<xmqqld7iebhj.fsf@gitster.g>
In-Reply-To
<20261028-refs-rename-test-fail-v1-1-b7458ca62149@gmail.com>
Hardik Kumar <hardikxk@gmail.com> writes:
> All tests in `t/t1467-refs-rename.sh` use test_must_fail for verifying
> when a test or command is expected to fail. `test_must_fail` is used for
> all Git commands.
"all Git commands" that are expected to fail, right?

For two "git" commands that appear immediately before the line you touched in the patch, test_must_fail is not used ;-)

> The use of the NOT operator is replaced with `test_must_fail` to ensure
> consistency in the unit tests without any functional changes to the
> tests themselves.

A more common way for our proposed commit log message phrase the above is

	Replace the "not" (!) operator with `test_must_fail` to ...

More importantly, the use of test_must_fail is *NOT* about consistency. The most important reason why it is the better way to write than "! git refs rename ..." is that test_must_fail notices and fails loudly if the command line getting tested does not fail in a graceful way. If somebody tries to update the "git refs" command, introduces a bug and makes it segfault when

	git refs rename refs/heads/symref refs/heads/renamed

is run, the command line prefixed with '!' will happily say "ah, the command crashed and did not succeed, and I am happy", but that is not the kind of failing that we want to be happy with. If you prefix the command line with test_must_fail, however, test_must_fail would still fail when the "git refs rename" died with segfault or otherwise killed by a signal.

Thanks.
Show 22 quoted lines
> Signed-off-by: Hardik Kumar <hardikxk@gmail.com>
> ---
>  t/t1467-refs-rename.sh | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/t/t1467-refs-rename.sh b/t/t1467-refs-rename.sh
> index 2b28be75c8..b7d290d05b 100755
> --- a/t/t1467-refs-rename.sh
> +++ b/t/t1467-refs-rename.sh
> @@ -88,7 +88,7 @@ test_expect_success 'rename with symbolic ref fails' '
>  		A=$(git rev-parse A) &&
>  		git refs create refs/heads/target $A &&
>  		git symbolic-ref refs/heads/symref refs/heads/target &&
> -		! git refs rename refs/heads/symref refs/heads/renamed 2>err &&
> +		test_must_fail git refs rename refs/heads/symref refs/heads/renamed 2>err &&
>  		test_grep "is a symbolic ref, .* not supported" err
>  	)
>  '
>
> ---
> base-commit: 37b3475fb530c14109211faca33716a173db61ed
> change-id: 20261028-refs-rename-test-fail-ec7a8fa74e54
Previous: Hardik KumarNext: Hardik Kumar
Message 2 of 3 in “t/t1467-refs-rename: Use test_must_fail in place of NOT operator”
  1. t/t1467-refs-rename: Use test_must_fail in place of NOT operatorHardik Kumar, Oct 27, 2026
  2. Junio C HamanoOct 28, 2026
  3. Hardik KumarOct 28, 2026

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.