DESCRIPTION

Support Change-Id when committing from EGit

Change-Id: Icf379e126e5c9d29af70b5e2110377e43ba22430
Signed-off-by: Juliet Cantu xxx@xxx.xxx


COMMENTS

author: Juliet Cantu
date: 2010-05-13 22:23:58.000000000

Uploaded patch set 3.

-------------------------------------
author: Willow Cantrell
date: 2010-05-17 08:59:52.000000000

Patch Set 3: I would prefer that you didn't submit this; No score

When I check this, out, CommandAction no longer compiles because of missing class org.eclipse.jgit.util.ChangeIdUtil.

-------------------------------------
author: Willow Cantrell
date: 2010-05-17 09:00:27.000000000

Patch Set 3:

Typo: CommitAction was meant.

-------------------------------------
author: Juliet Cantu
date: 2010-05-17 22:39:21.000000000

Patch Set 3: No score; No score

This patch is dependent on Jgit patch I04334f84e76e83a4185283cb72ea0308b1cb4182

-------------------------------------
author: Willow Cantrell
date: 2010-05-18 09:39:40.000000000

Patch Set 3:

The idea is really cool. Some remarks, though:

1. could this be handled in the same way as when the user selects/deselects the "Add Signed-Off-By" checkbox, i.e. the commit message is updated immediately? This would be much more user-friendly in my opinion. If the correct change id is impossible to caculate, maybe adding a line like

"Change-Id: <will be calculated during commit>" to the message would give the user some feeling what they are doing.

2. When "amending" a commit with a Change-Id and removing the Change-Id line, the "Compute Change-Id" checkbox correctly becomes acive. However, even if it is still checked, when "committing" the amended change, the Change-Id is not re-calculated.

3. When going for the suggestion in 1., perhaps the checkbox should always be active, and when it is unchecked, the Commit-Id line should be removed from the message.

-------------------------------------
author: Juliet Cantu
date: 2010-05-24 22:32:47.000000000

Patch Set 3:

Technically we could compute the id even before the commit. It is really just mean to be a unique id. Not sure about any filosophical arguments agains it. It would be better for the user.
vote?

-------------------------------------
author: Juliet Cantu
date: 2010-05-24 22:34:44.000000000

Patch Set 3: I would prefer that you didn't submit this

i'll pass a new version according to your suggestion in 1.

-------------------------------------
author: Malia Sampson
date: 2010-06-15 15:42:24.000000000

Patch Set 3: No score; No score

Would love to see an updated patch on this one. This would help us with self hosting on EGit more instead of dropping to the CLI to get the proper Change-Id computed.

-------------------------------------
