Re: [PATCHv2 0/3] Improve branch UI for setting upstream information

> Junio C Hamano <gitster@pobox.com> writes:
>
>> Carlos Martín Nieto <cmn@elego.de> writes:
>>
>>> As a result of making --unset-upstream fail if the given branch
>>> doesn't exist, I discovered a copy-paste error in on the the tests in
>>> the patch after it, so I'm resending the whole thing.
>>>
>>> The changes from the last reroll are the tightening of the situations
>>> where git will show an error message (not it's just if the branch is
>>> new and exists as remote-tracking) which I already sent as a reply in
>>> the other thread; and making --unset-upstream error out on bad input,
>>> which I already mentioned above.
>>
>> Thanks.
>>
>> In addition to "--unset-upstream must fail on i-dont-exist branch"
>> in [2/3], I am wondering if we would want to also make sure the
>> command fails when the upstream information is not set for the
>> branch, i.e. something like the following on top.
>>
>> What do you think?
>>
>>  t/t3200-branch.sh | 4 +++-
>>  1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git i/t/t3200-branch.sh w/t/t3200-branch.sh
>> index 1018e8b..a0aaedd 100755
>> --- i/t/t3200-branch.sh
>> +++ w/t/t3200-branch.sh
>> @@ -393,7 +393,9 @@ test_expect_success 'test --unset-upstream on HEAD' \
>>       git branch --set-upstream-to my14 &&
>>       git branch --unset-upstream &&
>>       test_must_fail git config branch.master.remote &&
>> -     test_must_fail git config branch.master.merge'
>> +     test_must_fail git config branch.master.merge &&
>> +     test_must_fail git branch --unset-upstream
>> +'
>
> Yeah, this looks good, makes sure that it will still behave correctly
> even if the code path for these two situations diverges.