FYC: migration of AMIA Open Source repositories from GitHub to alternative hosting

[EG-tech]

Greetings!

I recognize that this would be a significant undertaking given the amount of time/effort that has been invested by the AMIA Open Source committee and its various members/contributors/maintainers in using GitHub to host all our various repositories and tools. But - speaking as the maintainer of one such repository! - I have grown increasingly uncomfortable over the years since I started the Cable Bible and used GitHub Pages to host it with using GitHub as the working platform of either my personal or professional work. In retrospect, there has always been something of a disconnect in using a closed-source platform to host open-source work, and the acquisition by Microsoft hammered home (to me) some of the unpleasant ethical implications of that arrangement. Though I've generally been content to move most of my own day-to-day work to other options (like GitLab) and set up mirrors on GitHub to allow for continued discovery/use by the many archival professionals I know are on here, moves from the last couple of years like contracting with ICE and commercializing a proprietary machine-learning platform that likely violates copyleft licenses has inspired me to, at the very least, finally wonder out loud what the rest of the AMIA Open Source community thinks about the situation and whether it is worth investigating/investing in a migration to another host.

Ideally I'd raise this question e.g. at an in-person AMIA conference or Open Source Committee meeting, but given such a decision would have broad-reaching effects for many, far-flung contributors (many of whom may no longer be or were never AMIA members!) it also seems worth having the discussion here or in a similar agreed-upon virtual space.

In fact, as I get this far, realizing the gravity/effort that such a proposal implies, I'd just as soon start with the procedural question: if I wanted to propose a migration of either:

1. the Cable Bible to an alternative host like GitLab or Codeberg Pages
2. all AMIA Open Source repositories to the same or similar platform (volunteering myself to contribute significant time to such an effort, of course! 😃 )

...how might I formalize such a proposal to the Committee and how would it be considered?

All the best, y'all!

[dericed]

Hi @EG-tech, thanks for sharing all this and for volunteering some time to experiment on alternatives. We don't have an amiaopensource acct in github yet, but would you consider moving a copy of the cable-bible repo there and sharing the experience back. We could then repoint the homebrew installer to the gitlab url and see how this goes.

[EG-tech]

For sure! I could even try it with a couple options. That makes sense as a starting point, thanks @dericed

[retokromer]

FYI: Installing resources located elsewhere than GitHub works very well with Homebrew.

[EG-tech]

Successfully copied the Cable Bible repository to both GitLab and Codeberg, testing that the live site could be published on their Pages alternatives as well. For reference:

GitLab:
https://gitlab.com/EG-tech/cable-bible/
https://eg-tech.gitlab.io/cable-bible

Codeberg:
https://codeberg.org/eg-tech/cable-bible
https://eg-tech.codeberg.page/cable-bible/

A few thoughts:

• Both platforms offer automated import from an existing GitHub repository as an option when making a new repo, the only requirement being to generate a Personal Access Token in the GitHub account that owns the repo. This was quire simple and I assume could be performed by anyone with ownership access to the amiaopensource organization account.
• Both platforms offer organizational accounts, so could create equivalents to the amiaopensource organizational account
• With both, though the code and commit history copied over with seemingly no problem, Issues and Tag/Release history did not copy over. This may have had to do with permissions/scope of the Personal Access Token I selected when generating in GitHub, or it may be a broader limitation. I will investigate further, I imagine losing Issue Tracker history would be of some concern (it would be something I would like to move over with the Cable Bible if at all possible, for instance)
• the Pages process (hosting a live site, relevant to repos like Cable Bible and ffmprovisr) was dead simple on Codeberg, which just automatically creates a live page if there is an index.html file at the top layer of the repo, no further action required. However, it doesn't have any process/CI for generating a live page using a Static Site Generator as GitHub and GitLab do. That seems to be in the roadmap, but in general Codeberg is a newer service and probably has some other service limitations here or there.
• GitLab meanwhile has a slightly more complicated Pages process than GitHub, as you need to learn a bit about generating valid .gitlab-ci.yaml files, even if you have pre-built HTML ready.

On the whole, the process of moving over code/repositories is extremely simple. The more tricky bit is/would be maintaining contribution workflows - account access, ownership, issue and release history, etc.

I've tagged test releases in both repos above that can be used to test installation via Homebrew, but to Reto's point that all seems very much the same and is pretty much just a matter of adjusting a URL in the Homebrew formula.

[retokromer]

@EG-tech Are you attending the AMIA conference? This might be a good possibility to discuss whether AMIA will stay on GitHub or move to GitLab (or whatever).

[retokromer]

See also this article about Microsoft using an AI software that scans GitHub for interesting algorithms without the consent of the authors.