Gerrit Auto Cleanup: Difference between revisions

Latest revision as of 15:49, 25 November 2016

My change got abandoned by something calling itself "Qt Cleanup Bot", with a comment pointing me here! What's going on?!

To keep our dashboards useful, we have an automatic cleanup system that abandons changes which appear to be stale (as determined by seeing no activity at all within three months).

What do you care for my dashboard?! Just remove yourself from these changes!

It's not just your dashboard. Gerrit is common infrastructure. The most efficient use of that infrastructure is facilitated by unified workflows. Among other things, this means that the Gerrit status of changes is supposed to reflect their actual state.

Asking reviewers to simply "unsubscribe" is the wrong approach, because then they will miss any followup activity unless you explicitly re-invite them. It's not up to you to decide who ought to be interested in your changes; many reviewers (in particular maintainers) use Gerrit watches to monitor incoming changes in their areas of interest.

You still have no right to destroy my work like that!

Nothing is destroyed. It's two changed bytes in a database.

Fine. What now?

Your next action depends on the situation:

You are not actually reading this, because you disappeared without a trace. Don't do anything. Obviously.
You agree with the bot's judgement. Don't do anything. Unless you feel like explaining yourself.
- If you change your mind later, use reviewer:"Qt Cleanup Bot" to find a cleaned up change.
You think the change is still relevant, but realistically speaking, you are not going to work on it anytime soon. Hit the "Defer Change" button to signal your intent. The change will become easily reachable on your dashboard, but without cluttering the working set. When you finally get to work on the change again, use the "Restore Change" button.
You were really just about to update the change. Hit the "Restore Change" button directly.
You were actually waiting for the change to be reviewed/integrated, and are mightily annoyed now. Don't despair, and hit the "Restore Change" button to get the process going again. Things to check:
- Did you add reviewers?
  - Is the code relevant to them?
  - Are they still active?
  - Did you try to remind them of the pending review?
- Did you adequately respond to previous comments?
- Maybe the change was already approved and you just forgot to stage/submit it? Or to try again after an integration failure due to an unrelated change or test instability?

@@ Line 1: / Line 1: @@
-====My change got abandoned by something calling itself “Qt Cleanup Bot”, with a comment pointing me here! What’s going on?!====
+[[Category:Tools::Gerrit]]
+__NOTOC__
+==== My change got abandoned by something calling itself "Qt Cleanup Bot", with a comment pointing me here! What's going on?! ====
 To keep our dashboards useful, we have an automatic cleanup system that abandons changes which appear to be stale (as determined by seeing no activity at all within three months).
-====What do you care for '''my''' dashboard?! Just remove yourself from these changes!====
+==== What do you care for '''my''' dashboard?! Just remove yourself from these changes! ====
-It’s not just your dashboard. Gerrit is common infrastructure. The most efficient use of that infrastructure is facilitated by unified workflows. Among other things, this means that the Gerrit status of changes is supposed to reflect their actual state.
+It's not just your dashboard. Gerrit is common infrastructure. The most efficient use of that infrastructure is facilitated by unified workflows. Among other things, this means that the Gerrit status of changes is supposed to reflect their actual state.
-Asking reviewers to simply “unsubscribe” is the wrong approach, because then they will miss any followup activity unless you explicitly re-invite them. It’s not up to you to decide who ought to be interested in your changes; many reviewers (in particular maintainers) use Gerrit watches to monitor incoming changes in their areas of interest.
+Asking reviewers to simply "unsubscribe" is the wrong approach, because then they will miss any followup activity unless you explicitly re-invite them. It's not up to you to decide who ought to be interested in your changes; many reviewers (in particular maintainers) use Gerrit watches to monitor incoming changes in their areas of interest.
-====You still have no right to destroy my work like that!====
+==== You still have no right to destroy my work like that! ====
-Nothing is destroyed. It’s two changed bytes in a database.
+Nothing is destroyed. It's two changed bytes in a database.
-====Fine. What now?====
+==== Fine. What now? ====
 Your next action depends on the situation:
+* You are not actually reading this, because you disappeared without a trace. Don't do anything. Obviously.
-* You are not actually reading this, because you disappeared without a trace. Don’t do anything. Obviously.
+* You agree with the bot's judgement. Don't do anything. Unless you feel like explaining yourself.
-* You agree with the bot’s judgement. Don’t do anything. Unless you feel like explaining yourself.
+** If you change your mind later, use reviewer:"Qt Cleanup Bot" to find a cleaned up change.
-** If you change your mind later, use reviewer:“Qt Cleanup Bot” to find a cleaned up change.
+* You think the change is still relevant, but realistically speaking, you are not going to work on it anytime soon. Hit the "Defer Change" button to signal your intent. The change will become easily reachable on your dashboard, but without cluttering the working set. When you finally get to work on the change again, use the "Restore Change" button.
-* You think the change is still relevant, but realistically speaking, you are not going to work on it anytime soon. Hit the “Defer Change” button to signal your intent. The change will become easily reachable on your dashboard, but without cluttering the working set. When you finally get to work on the change again, use the “Restore Change” button.
+* You were '''really''' just about to update the change. Hit the "Restore Change" button directly.
-* You were '''really''' just about to update the change. Hit the “Restore Change” button directly.
+* You were actually waiting for the change to be reviewed/integrated, and are mightily annoyed now. Don't despair, and hit the "Restore Change" button to get the process going again. Things to check:
-* You were actually waiting for the change to be reviewed/integrated, and are mightily annoyed now. Don’t despair, and hit the “Restore Change” button to get the process going again. Things to check:
 ** Did you add reviewers?
 *** Is the code relevant to them?
@@ Line 29: / Line 31: @@
 ** Did you adequately respond to previous comments?
 ** Maybe the change was already approved and you just forgot to stage/submit it? Or to try again after an integration failure due to an unrelated change or test instability?
-If you are a reviewer and want to adopt the change, you can ask the owner or an admin to hit the “Defer/Restore Change” button for you. Afterwards, you can just push updated patchsets into the foreign change.