[mhitza]

It's not only discoverability, but also the need for accounts.

At some point I made very tiny contributions to OSS projects that had their own Gitlab hosted instances.

Sure a password manager makes it tolerable, but what about having an anonymous way of opening up PRs (subject to owner moderation of course)?

Use the author name and email for a virtual identity and when the PR request is accepted (not merged) force an email address validation for the PRer so that comment interaction can happen via email.

[xigoi]

This is an entirely self-inflicted problem. Git was designed so you can just send someone a patch by e-mail. It was GitHub that introduced “pull requests”, which require an account, as the main method of collaboration. Vendor lock-in as a feature.

[freeopinion]

Or...

Fork the original repo onto your forge. Your forge could be your own forgejo instance, your own GitLab instance, just a plain git repo, or even Github if you like it so much.

Then send a note to the original authors informing them of your fork and the patch you wrote. Request that they review your work and pull it into their repo if they approve of it.

You don't need an account on their forge and they don't need an account on your forge.

[mhitza]

For me that adds to many steps, and removes collaboration transparency.

Disregarding the fact that I would need to have a way to reliably find the contact information, the review process would probably take place outside the source code platform.

[freeopinion]

Too many steps? How is it more steps than submitting an MR or PR today? You have to fork, code a patch, submit a pull request. The difference is just that it happens across forge instances.

You have a point about the review process. How much of a moat do you want around your project? You want to keep spammers out. But that creates hurdles.

I was trying to interact with the Trunk project the other day. But they required me to create accounts on multiple services just to get in the game. When you need a Github account and a Discord account and... You know what? Nevermind. It's too much effort just to be nice and inform you of an issue with your project.

I totally understand that putting a public email address out there creates a huge burden for them. I suppose if I value a product enough I will accept some smaller burden and take the load off them. But most of the time I don't value a product enough to allow the authors to offload their burden onto me. Conversely, they don't value my input enough to accept the huge burden it would take them to allow me to interact with them.

[freeopinion]

What if you could submit a PR by pasting the url of a repo commit?

The forge could automatically figure out if the submitted repo was a fork, figure out the sequence of commits in the foreign repo that diverged from itself, analyze the commit comments and commit diffs to see if they jived, and add the PR if it passed the spam filter.

Then you could submit PRs without an account or even having the contact information of a maintainer.