[INFRA-2697] Unfork repos #2272
Comments
|
This is a work in progress I've recently started with jenkins-infra/infra-reports#24 and jenkins-infra/jenkins.io#3585. I plan to send a proposal to the dev list this week. In another issue tracker this is known as JENSEC-921. |
|
Daniel Beck I see your PRs have been merged, is there still something to do for this issue or can it be closed? |
|
Still open. |
|
Could you consider unforking of these repos please? |
|
cc @MarkEWaite, @jmMeessen, @res0nance ^^ |
|
I've got one of these too - https://github.com/jenkinsci/jakarta-mail-api-plugin - but I have no idea how I would go about unforking it. |
|
If nothing changed over the past couple of years, this is still done manually by submitting a request to GitHub through its bot assistent. |
|
Based on https://stackoverflow.com/questions/38831301/how-to-un-fork-the-github-repository/66470086#66470086 I was able to use their virtual assistant to submit a request to unfork the elastic axis plugin and the git plugin. I hit the rate limit when I tried to unfork the node label parameter plugin |
|
I used the virtual assistant recently for one, plugin-pom I think. It was done fairly quickly |
All four of my requests are now complete. The request for git plugin is resolved. |
|
Btw, is there a policy for the forking process, because https://github.com/jenkinsci/ecu-test-execution-plugin was just forked ~10h ago and may be considered as unforking candidate sooner or later 🤔 |
|
see jenkins-infra/repository-permissions-updater#2679 (comment) (wording could probably be updated as there's now an official process from GitHub side on it) |
|
I successfully used the virtual assistant to unfork https://github.com/jenkinsci/jakarta-mail-api-plugin - thanks! |
|
I submited a request for repositories I can manage 👍🏻 Edit: Btw, basil, where did you get this list from? If I query forks via https://github.com/orgs/jenkinsci/repositories?q=&type=fork&language=&sort= I receive 776 results 👀 |
|
Just browsing around a few popular plugins. |
|
There's a report here that has all plugins in this situation: |
|
I've created a request for the snakeyaml one |
If we want to continue detaching forks, I'd recommend that one person submits all eligible repositories on behalf of the jenkinsci umbrella organization. |
|
One of the questions is 'have you committed to the repository' I don't think one person will meet that criteria, although possibly they would let an organisation owner do it. |
That was no blocker for a repo I detached an hour ago.
I just asked them on my ticket, 'cause I'd rather not want to submit 770 tickets. |
dduportal
modified the milestones:
infra-team-sync-2022-08-09,
infra-team-sync-2022-08-16
Still ongoing. |
|
GH evaluated the feasibility internally, but given unforking is a manual resource and time intensive process, GH is unable to unfork the vast amount of forks we host. However, they offered us to change all fork network owners to the jenkinsci organization, if desired. |
|
Should be fine to do that, cc @daniel-beck |
|
I sent you the email to authorize the process, Tim. I could create a forum post later on, if we want to proceed with this. |
This would be wrong for some repos, so we would provide either a list of repos to apply this to, or a list of repos to not apply this to. Otherwise https://github.com/jenkinsci/jgit https://github.com/jenkinsci/rubywm https://github.com/jenkinsci/plexus-utils https://github.com/jenkinsci/jna https://github.com/jenkinsci/jmdns-fork would probably upset some folks. |
FTR https://www.jenkins.io/doc/developer/publishing/source-code-hosting/forks/ |
|
We could delete those forks? I don't think any of them are used now |
These are a handful of examples I found when browing through the repos list. Is that something you can confidently say about every fork that should be a fork? We'll need this list of forks to change the root of and forks to keep as forks anyway, to be able to delete the latter, so we do not need to provide a list to GH. |
I started writing down a non exhaustive list of forks that should remain as is, given those are legit forks outside the plugin hosting process: https://gist.github.com/NotMyFault/dfec71b094419f45145d2c187636b6b1 |
Can you do this as a list of all forks we have, and label them "actual fork" and "hosting fork"? (FWIW I'd use a 2 column Google Doc for this.) That way it's easy to confirm all repos were looked at, and an explicit decision to unfork/root (or not) was made. Especially interesting also for weird edge cases (we also have some non-plugin forks of vaguely Jenkins related repos of contributors…) |
|
GitHub have not offered to process a list from what I understand they would do all of them or small numbers. If we don’t want something unforked and we want to do this process we need to delete the repo, which could mean moving it to a temp account and back |
Right, but that still means we should have a complete list to minimize the risk that we have forked another project and suddenly become root of their network. |
Right, GH's offer covers all forks.
Based on the fork report, I generated https://docs.google.com/spreadsheets/d/197A5yzVCfBsXkRF3Uh9JbSUKD2kU7q-z6O8rNxLdhNE/edit#gid=0, which highlights repositories that should remain as is and we would need to move to a temp account (or possibly delete, if no commits have been made and the repository brings in nothing of use and and value). |
Right, and now explicitly mark those that should have their network root in jenkinsci. TBH I'd be pretty surprised if that's trivially decidable for everything not currently marked to remain a fork. |
|
I've reviewed all of the repositories not containing
I don't quite understand. Shouldn't all the plugins have their network root as jenkinsci? |
All all repos not marked to remain a fork plugins? No, plenty of counter-examples. Are all So I suggest you go through the list and mark every entry with "Should remain fork", or "Is Jenkins plugin", as applicable. See whether anything remains unmarked. |
|
I seperated the list across different sheets on the spreadsheet, but I also didn't find anything new reading over it. Given that changing fork networks is a simple procedure on GH's end and we discover a fork we want to restore in the future, we can always go ahead and do so and prepare the few forks giving GH the go. |
|
The point of my recent messages is that we need to be able to distinguish "action should be taken" and "nobody has looked at this so we don't know yet". Yes, looking at 700+ repos to see whether they're Jenkins plugins, Jenkins related, or something else, is annoying, but it's IMO irresponsible to ask for batch action without any confidence that this won't mess with other folks' projects. I know of at least one repo currently marked for rooting in this doc that shouldn't be because it's an actively maintained open source project unrelated to Jenkins that we have a fork of. This alone tells me that the "Jenkins plugins" tab was not populated carefully (even assuming you intended it to be wider than Jenkins plugins). |
As the maintainer of https://github.com/jenkinsci/gearman-plugin.git I would very welcome it to become its own fork network (in case you are seeking feedback). The plugin development was originally done on OpenStack Gerrit instance with a mirror set to https://github.com/openstack-infra/gearman-plugin . I think I forked it under the jenkinsci organization for convenience but there is really no need to keep that fork relation since the parent was merely a mirror, it had no issue tracker and the PR were automatically rejected. When proposing a PR to Thanks for leading that effort with GitHub! |
|
@hashar Note that GH said they cannot disconnect our stuff in batch (#2272 (comment)), but instead offered to make the jenkinsci fork the new root of the network. |
As noted in cli/cli#1467 there are various repositories in @jenkinsci which are still formally considered forks of long-dormant personal repositories, even though they are in fact official. We should remove the linkage. jenkins-test-harness-htmlunit and plugin-pom I have noticed, but I am sure you could find others with a search.
Originally reported by jglick, imported from: Unfork repos
The text was updated successfully, but these errors were encountered: