GitHub or Codeberg?

[semuadmin]

There is growing concern in some developer quarters with regard to hosting open source projects on GitHub, e.g.

• Microsoft's corporate policies.
• The increasing intrusion of Copilot AI, potentially in violation of open source license attribution conditions.
• The implications of data being hosted under US jurisdiction.

While GitHub remains by far the largest source code host, with over five billion developer contributions made to more than 500 million open source projects in 2024, many are advocating a move to alternative platforms - in particular Codeberg.

I personally remain fairly open minded on this issue but am interested to understand the views of the wider user and contributor communities.

[gdt]

I wouldn't say that Copilot "potentially" violates licenses. LLMs trained on stolen data are an outright violatiion even if a BSD license because they omit attribution. So it is absolutely a violation. Further, people use these tools and submit LLM output, and that code if merged taints the license status of the project, as it was copied improperly (from the LLM training set). As for MS policies, I think the big thing is how github is being run, AI-in-your-face, training on github-hosted code.

I'm not sure what the issues are for open source code being hosted in the US. For any country, people worry about police/intelligence access to data, but all the data in a github project is already public. While I know people object to some US government policies, the EU is poised to ban e2e encryption, and the UK is trying to do that with iCloud, so my take is that there are going to be issues everywhere you turn.

All that said, I am 100% in favor of moving to Codeberg if you the main contirbutor and project lead can stand the disruption. While it's codeberg today, the forgejo code is Free and people could stand up their own, so I feel it's a much more stable future.