Most orgs like having a single source of truth for their projects, though (for instance, the Linux kernel GitHub repo is simply a mirror). In fact, if anything, this project could set up mirroring so repos are mirrored from GitLab to GitHub, etc.
But then, the Linux kernel really have multiple sources of truth, depending on what version you are talking about (Linus branch, mainline, vendor forks, etc)
The main point isn't about the source code, it's about all the rest : bug report, discussions, planning..
I run my own gitea instance, and for repos that are in multiple places, I leave a note in the readme as to what the canonical URL is. If someone wants to open a PR or an issue on Github I don't wanna be hostile by having them create another account just to ask a question. Once the patch is merged, it gets pushed on all remotes.
If you have a large project with lots of contributors and activity, and you actively want to discourage contributions in the wrong place, you can disable issues or PRs for that one repo.
Also, people browse code using GitHub clients, serving an empty Readme is not a good idea to break that behaviour.
As someone who runs a couple small projects with large userbases, having multiple issue queues across multiple platforms is a massive organizational hassle. Having one canonical place for issues makes life easier for all parties.
With that said, if you disable issues on a system, make sure you have prominent documentation that says "Submit issues to XXX". Otherwise, it's confusing.
And for the users. Given that it's fairly technical software, it's reasonable to expect that folks would search the issue queue before submitting an issue. If the issue queue was split across multiple sources, issues would exist in one system and not the other.
Sorry if you disagree, but giving clear instructions about how to report issues and limiting vectors is fair, logical, and incredibly common and a best practice.
If said casual user doesn't want to bother finding where to report issues (which a competent maintainer would surely document in the main README), there's also a fairly good chance it's not a big deal to begin with or that the explanation is vague/unactionable, at least from my personal experience.
Why aren't issues, etc., part of the repo and so pulled (optionally) to duplicate repos. Shouldn't web front ends be views of federated content -- having issues in one git web front-end not get propagated to others for the same project is surely a major flaw.
16 comments
[ 5.6 ms ] story [ 46.8 ms ] threadThe main point isn't about the source code, it's about all the rest : bug report, discussions, planning..
I run my own gitea instance, and for repos that are in multiple places, I leave a note in the readme as to what the canonical URL is. If someone wants to open a PR or an issue on Github I don't wanna be hostile by having them create another account just to ask a question. Once the patch is merged, it gets pushed on all remotes.
If you have a large project with lots of contributors and activity, and you actively want to discourage contributions in the wrong place, you can disable issues or PRs for that one repo.
Also, people browse code using GitHub clients, serving an empty Readme is not a good idea to break that behaviour.
But when I go to a project on Github to report an issue and I can't, then I can't help but think "well fuck you too."
With that said, if you disable issues on a system, make sure you have prominent documentation that says "Submit issues to XXX". Otherwise, it's confusing.
It makes life easier for you.
It doesn't make life easier for the casual user of your software who just wants to report one issue.
Sorry if you disagree, but giving clear instructions about how to report issues and limiting vectors is fair, logical, and incredibly common and a best practice.