Closed rhaschke closed 8 years ago
... we now have a kinetic version too because of C++11
I would have loved to use a squashing commit, i.e. forwarding jade-devel to the cherry-pick branch. @davetcoleman @v4hn Anybody minds to rebase jade-devel to there?
Hopefully this facilitates reading the git graph: Merge commits are unique contributions, while cherry-picks could be committed directly. I just had to use a PR to let Travis do the compile checking.
As nobody complained, I have reset jade-devel to 3f16bda. Please be careful to continue on the correct commit, when you already pulled the jade-devel update:
git checkout jade-devel git reset --hard origin/jade-devel
Anybody minds to rebase jade-devel to there?
Yes, please never rebase any of the official branches (unless someone decides to push 100MB+).
If you don't want to see a request merged by somebody else, please indicate that in the title/description.
I agree though that dave should have squash-merged though.
Sounds like a nice thing to discuss next week :)
I like the policy to merge once and squash/commit to other branches.
Sadly, this doesn't play nice with git cherry
though, so I'm unsure whether it's a good idea.
In this regard rebasing the request (with all its individual commits) on top of the other branches and fast-forwarding the branches seems to be the better choice...
...
Sorry, just did it. As long as nobody continued to work on that branch, there is no harm. That's why, I didn't want wait any longer. I've seen both of you being active and you didn't reply (in time) to my two requests.
I saw that much..
Please don't next time.
People (at least me) tend to freak out when wstool update
/ git pull
fails because someone decided to rebase official branches.
Yea, I agree we shouldn't ever force-push the main ros-planning repos, which is what I'm assuming you just did. I'm also not clear why you didn't cherry-pick the necessary commits into a new branch for the target ROS-version and create a PR based on those cherry-picks - isn't that what you did? Similar to how I did it here. Please don't create PRs claiming to do something I desire (cherry-pick to newer branch) but isn't labeled "do not merge".
Though I'm a bit confused by what you guys mean by squash-merged.
I'm also not clear why you didn't cherry-pick the necessary commits into a new branch for the target ROS-version and create a PR based on those cherry-picks - isn't that what you did?
I didn't cherry-picked all commits individually, but rebased and squashed:
git rebase -i --onto <new branch> <base commit> <old branch>
To facilitate this task, github allows to "squash and merge". Simply click the option in the dropbox of the "Confirm merge" button.
Though I'm a bit confused by what you guys mean by squash-merged.
That's what we mean ;-)
Interesting - I've never used that feature of Github!
Seems like a superior option for keeping the git history clean - are there use cases where we would want to do a regular merge? Or should we just disable that option for all the git repos?
As said, I suggest to use
On Mon, Jul 18, 2016 at 11:28:17PM -0700, Robert Haschke wrote:
As said, I suggest to use
- regular merge commits whenever a genuine contribution (with several commits) is going to be merged. The merge directly visualizes this fact in the git history graph.
- squash merges whenever there is only a single commit to be merged, or when we cherry-pick a set of commits from another branch. This indeed keeps the history clean.
Let's simply discuss this during the next meeting. :)