Re: What's cooking in git.git (Aug 2016, #01; Tue, 2)

Hi Lars & John,

On Thu, 4 Aug 2016, Lars Schneider wrote:

quoted

On 04 Aug 2016, at 13:32, John Keeping [off-list ref] wrote:

On Thu, Aug 04, 2016 at 10:03:39AM +0200, Lars Schneider wrote:

quoted

quoted

* jk/push-force-with-lease-creation (2016-07-26) 3 commits
- push: allow pushing new branches with --force-with-lease
- push: add shorthand for --force-with-lease branch creation
- Documentation/git-push: fix placeholder formatting

"git push --force-with-lease" already had enough logic to allow
ensuring that such a push results in creation of a ref (i.e. the
receiving end did not have another push from sideways that would be
discarded by our force-pushing), but didn't expose this possibility
to the users.  It does so now.

Will merge to 'next'.

t5533-push-cas.sh "16 - new branch already exists" seems to be broken 
for OSX on next. Git bisect indicates that "push: add shorthand for 
--force-with-lease branch creation" might be the culprit.

https://travis-ci.org/git/git/jobs/149614431
https://api.travis-ci.org/jobs/149614431/log.txt?deansi=true (non-JS)

It seems that the test script has already done "test_commit C", so the
newly added "test_commit c" does nothing on a case-insensitive
filesystem.

Something like this will make the test more consistent with the rest of
the file:

diff --git a/t/t5533-push-cas.sh b/t/t5533-push-cas.sh
index 5f29664..e5bbbd8 100755
--- a/t/t5533-push-cas.sh
+++ b/t/t5533-push-cas.sh

@@ -220,7 +220,7 @@ test_expect_success 'new branch already exists' '

	(
		cd src &&
		git checkout -b branch master &&
-		test_commit c
+		test_commit F
	) &&
	(
		cd dst &&

Confirmed. This patch fixes the issue!

Funny. I worked heads-down to have some kind of Continuous Integration to
run on my laptop, and this breakage came up. I fixed it locally, and only
then did it occur to me that it might have been fixed already, and then I
found this mail with a patch identical to mine.

Will send out the patch in a moment.

Thanks,
Dscho