1. 17

  2. 3

    I find it facanating and scary that Github has managed to attract a number of large projects to move over. In python’s case though I wonder if there was a github equivalent for hg would they have moved to that over GitHub?

    1. 5

      BitBucket? Or if you wish to host-your-own, Kallithea?

      1. 5

        I don’t think so.

        There were some discussions a while back, posted here or HN, about moving Python away from Mercurial to Git. I don’t recall GitHub being mentioned at the time, but there seemed to be a strong desire to move to Git. The biggest issue was that most contributors said they were more comfortable with Git, but I think there may have been other issues with Mercurial.

        I’m also hesitant about everything moving to GitHub, but have to admit it’s a pretty great tool.

        1. 1

          I don’t recall any technical objections to hg itself. Well, there was this, which is a response to a bunch of uninformed opinions about what hg can or cannot do.

          It’s the great tragedy of hg: since it’s not popular, its features are not well-known, and its critics argue strongly from a position of unfamiliarity with it. Thus, it keeps viciously cycling deeper into obscurity. ;_;

        2. 2

          Probably not. They are moving to git+github because their contributors are already more familiar with these tools+platforms than hg.

          1. 2

            Not so much git as github. They just want whatever is popular because they want more contributions. They also don’t want to sysadmin themselves. That’s why gitlab and bitbucket were discarded.

          2. 2

            Curious, why does this scare you? Generally I agree, less centralization. But in this specific case, it’s sort of easy enough to move away if something were to happen, GitHub has showed little signs of being evil (at least that I’m aware of), and has overall been a great thing for open source.

            And as a plus, like others have mentioned, many are familiar with GitHub, so maybe their contributions will increase.

            1. 5

              I think it is not always quite so easy to move away. Lots of links and with no way to redirect them after a move.

              1. 5

                Yes: it’s not just the code, but the trail of issues, comments, and pull requests alongside that give it meaning. When the Cooper Hewitt Smithsonian Design Museum acquired the application Planetary in 2013, they got the whole repo transferred to them:

                Although Bloom folded in 2012, its three principals have not only gifted the code for Planetary to Cooper­-Hewitt they have also given us explicit permission to publicly release the source code under an open source (BSD) license, and its graphical assets under a Creative Commons (non-commercial) license.

                As a research institution we are also interested in reaching new understandings of the ways designers use code that can be gleaned from the code itself.

                As we are acquiring a source code from the version control system that it was managed in (also GitHub), we have been able to preserve all the documentation of bugs, feature additions, and code changes throughout Planetary’s life. This offers many new interpretive opportunities and reveals many of the decisions made by the designers in creating the application.

              2. 4

                I’ve personally found that once a repo is on GitHub and using their pull request system to manage changes, it becomes incomprehensible outside of GitHub. The history and discussion there ends up trapped within their interface. I’m also concerned that GitHub is replacing (or has replaced, depending on your view) SourceForge as “the” place to host code without any meaningful competition. Looking back at what happened with SourceForge, i’m slightly concerned we may be heading down the same path.