Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [cross-project-issues-dev] Question simrel and Gerrit use ...

On Thu, Sep 3, 2015 at 12:06 AM, David M Williams <david_williams@xxxxxxxxxx> wrote:
What "refs" do others use then submitting changes for "Gerrit review" for Mars_maintenance?
I've heard refs/for/review should work .. but after some initial trouble with that, I've used refs/for/Mars_maintenance.
And that seems to work. Am I doing it wrong? Or ... is refs/for/review wrong? Or, does it not matter?
(my initial troubles might have been due to something else, for all I know).

Pushing a commit to "refs/for/x" creates a review for this commit targeting branch "x", so if you are
pushing to "refs/for/review" the target branch would be "review". Pushing to refs/for/review doesn't
make sense if the branch "review doesn't exist and should be rejected by Gerrit.
When you submit the change it will be merged to it's target branch.

If you push the same commit for two different target branches you'll end-up
with two changes in review in Gerrit even if they have the same changeId. If later you push
another commit (which is usually created by amending the first one to react on review comments)
with the same changeId to the same target branch (refs/for/<target branch>) you'll get a new patchset
for the corresponding change in review.
 
A smaller issue, I notice that JWT has https://git.eclipse.org/r/#/c/48872/
out there, that was never merged, but it is 3 months old, so assumed it was merged the old-fashioned way?
Such old, "un-used" Gerrit commits should be "abandoned". (right?)
If for no other reason, makes my "watched" queries cleaner :)

right, stale or bad changes in review which aren't needed anymore should be abandoned to clean up the
queue of changes pending in review. Some projects automatically abandon changes which weren't updated
for e.g. 3 months. If required abandoned changes can be restored, so you won't loose a change if it
was abandoned by mistake. 

-Matthias

Back to the top