diff mbox series

[1/2] checkout: add tests for -b and --track

Message ID ebe696b2-bf68-ccd1-b930-dce274ec9279@web.de (mailing list archive)
State New, archived
Headers show
Series [1/2] checkout: add tests for -b and --track | expand

Commit Message

René Scharfe May 24, 2020, 7:22 a.m. UTC
Test git checkout -b with and without --track and demonstrate unexpected
error messages when it's given an extra (i.e. unsupported) path
argument.  In both cases it reports:

   $ git checkout -b foo origin/master bar
   fatal: 'bar' is not a commit and a branch 'foo' cannot be created from it

The problem is that the start point we gave for the new branch is
"origin/master" and "bar" is just some extra argument -- it could even
be a valid commit, which would make the message even more confusing.  We
have more fitting error messages in git commit, but get confused; use
the text of the rights ones in the tests.

Reported-by: Dana Dahlstrom <dahlstrom@google.com>
Original-test-by: Jeff King <peff@peff.net>
Signed-off-by: René Scharfe <l.s.r@web.de>
---
 t/t2018-checkout-branch.sh | 10 ++++++++++
 t/t2027-checkout-track.sh  | 24 ++++++++++++++++++++++++
 2 files changed, 34 insertions(+)
 create mode 100755 t/t2027-checkout-track.sh

--
2.26.2

Comments

Jeff King May 27, 2020, 6:40 a.m. UTC | #1
On Sun, May 24, 2020 at 09:22:51AM +0200, René Scharfe wrote:

> Test git checkout -b with and without --track and demonstrate unexpected
> error messages when it's given an extra (i.e. unsupported) path
> argument.  In both cases it reports:
> 
>    $ git checkout -b foo origin/master bar
>    fatal: 'bar' is not a commit and a branch 'foo' cannot be created from it
> 
> The problem is that the start point we gave for the new branch is
> "origin/master" and "bar" is just some extra argument -- it could even
> be a valid commit, which would make the message even more confusing.  We
> have more fitting error messages in git commit, but get confused; use
> the text of the rights ones in the tests.

Did you mean "more fitting error message in git checkout"?

> Original-test-by: Jeff King <peff@peff.net>

I didn't think I really contributed much, but OK. :)

> diff --git a/t/t2018-checkout-branch.sh b/t/t2018-checkout-branch.sh
> index 21583154d8..d99b699396 100755
> --- a/t/t2018-checkout-branch.sh
> +++ b/t/t2018-checkout-branch.sh
> @@ -260,4 +260,14 @@ test_expect_success 'checkout -b to a new branch preserves mergeable changes des
>  	test_cmp expect actual
>  '
> 
> +test_expect_success 'checkout -b rejects an invalid start point' '
> +	test_must_fail git checkout -b branch4 file1 2>err &&
> +	test_i18ngrep "is not a commit" err
> +'
> +
> +test_expect_failure 'checkout -b rejects an extra path argument' '
> +	test_must_fail git checkout -b branch5 branch1 file1 2>err &&
> +	test_i18ngrep "Cannot update paths and switch to branch" err
> +'

OK, covering the normal case without --track, both with and without the
extra arg. Makes sense.

> +test_expect_success 'checkout --track -b creates a new tracking branch' '
> +	git checkout --track -b branch1 master &&
> +	test $(git rev-parse --abbrev-ref HEAD) = branch1 &&
> +	test $(git config --get branch.branch1.remote) = . &&
> +	test $(git config --get branch.branch1.merge) = refs/heads/master
> +'
> +
> +test_expect_failure 'checkout --track -b rejects an extra path argument' '
> +	test_must_fail git checkout --track -b branch2 master one.t 2>err &&
> +	test_i18ngrep "cannot be used with updating paths" err
> +'

And these ones with --track, which produces a different error message.
Makes sense.

-Peff
René Scharfe May 28, 2020, 1:53 p.m. UTC | #2
Am 27.05.20 um 08:40 schrieb Jeff King:
> On Sun, May 24, 2020 at 09:22:51AM +0200, René Scharfe wrote:
>
>> Test git checkout -b with and without --track and demonstrate unexpected
>> error messages when it's given an extra (i.e. unsupported) path
>> argument.  In both cases it reports:
>>
>>    $ git checkout -b foo origin/master bar
>>    fatal: 'bar' is not a commit and a branch 'foo' cannot be created from it
>>
>> The problem is that the start point we gave for the new branch is
>> "origin/master" and "bar" is just some extra argument -- it could even
>> be a valid commit, which would make the message even more confusing.  We
>> have more fitting error messages in git commit, but get confused; use
>> the text of the rights ones in the tests.
>
> Did you mean "more fitting error message in git checkout"?

Dang, yes, a typo -- it's all about checkout.

René
diff mbox series

Patch

diff --git a/t/t2018-checkout-branch.sh b/t/t2018-checkout-branch.sh
index 21583154d8..d99b699396 100755
--- a/t/t2018-checkout-branch.sh
+++ b/t/t2018-checkout-branch.sh
@@ -260,4 +260,14 @@  test_expect_success 'checkout -b to a new branch preserves mergeable changes des
 	test_cmp expect actual
 '

+test_expect_success 'checkout -b rejects an invalid start point' '
+	test_must_fail git checkout -b branch4 file1 2>err &&
+	test_i18ngrep "is not a commit" err
+'
+
+test_expect_failure 'checkout -b rejects an extra path argument' '
+	test_must_fail git checkout -b branch5 branch1 file1 2>err &&
+	test_i18ngrep "Cannot update paths and switch to branch" err
+'
+
 test_done
diff --git a/t/t2027-checkout-track.sh b/t/t2027-checkout-track.sh
new file mode 100755
index 0000000000..d0b41d7cd0
--- /dev/null
+++ b/t/t2027-checkout-track.sh
@@ -0,0 +1,24 @@ 
+#!/bin/sh
+
+test_description='tests for git branch --track'
+
+. ./test-lib.sh
+
+test_expect_success 'setup' '
+	test_commit one &&
+	test_commit two
+'
+
+test_expect_success 'checkout --track -b creates a new tracking branch' '
+	git checkout --track -b branch1 master &&
+	test $(git rev-parse --abbrev-ref HEAD) = branch1 &&
+	test $(git config --get branch.branch1.remote) = . &&
+	test $(git config --get branch.branch1.merge) = refs/heads/master
+'
+
+test_expect_failure 'checkout --track -b rejects an extra path argument' '
+	test_must_fail git checkout --track -b branch2 master one.t 2>err &&
+	test_i18ngrep "cannot be used with updating paths" err
+'
+
+test_done