leah blogs: June 2018

05jun2018 · GitHub: quo vadis?

As GitHub user #139 I feel compelled to say something about GitHub getting bought by Microsoft.

I still remember, back when GitHub was founded, I was both thrilled and frightened. Thrilled, because the three founders managed to bootstrap a startup the right way: profitable from day one, focussed on a single, successful product, that did what people wanted. I told myself that if I’d ever do a startup, I’d do it like them. (Yet they took venture capital in 2012 and 2015 and as a result grew from 50 to over 800 employees by now, making loss at least in 2016 until they changed their business plans.)

Why was I frightened? I saw GitHub was immedately getting very successful, and many people, especially in the Ruby community, moved their projects to it, creating both a monopoly and a single point of failure.

One of my most successful projects, Rack, was converted from Darcs to Git in May 2008. I put it on GitHub (which was only a few months old by then) about that time, but I also provided my own Git mirror on own infrastructure. However, development quickly shifted to GitHub only, mostly because pull requests and issues were very convenient. Over time, my skepticism vanished, using GitHub was a no-brainer, and while occasional outages remembered us of the central position GitHub is in, we didn’t do anything.

Let me emphasize a few ways GitHub vastly improved my own open source work: finally, it was easy to report issues for many projects, without having to register yet another Bugzilla account, and issues could easily link issues at other projects. GitHub made it simple to quickly look into the actual source of many projects in a straight-forward way, without having to figure out CVS checkouts or fetch tarballs. It was easy to see which people contribute to which projects, and I discovered some cool projects this way.

So, now they are getting bought by Microsoft. I’m sad that they are getting bought at all, because I think it’s very important that such a central piece of the open source community stays independent of major software vendors. As for getting bought by Microsoft, I cannot share the enthusiasm many have: it is still a huge company that makes its profits primarily from proprietary, closed-source software and vendor lock-in, and while their management certainly changed a lot in the last decade, who knows how long this will last. Worse buyers are easily imaginable, however.

It is therefore sensible to think of alternatives to GitHub. Contrary to many, I don’t think switching to alternative offerings such as GitLab.com, BitBucket or SourceForge significantly improves the situation: while GitHub’s monopoly could get whittled down, we are still dependent on another for-profit company that is likely to be acquired by a major corporation sooner or later, too.

As for my own projects, I plan to be moving the most part of my recent projects (so called leahutils) from GitHub to a self-hosted solution. The details are not fixed yet, but I have enough experience with GitLab that I’m sure I’ll use something else. For these projects, I also have different needs compared to what GitHub offered: I’m often the sole committer, and I prefer receiving patches by mail and refining them myself rather than telling people how to improve their pull requests. So likely, I’ll set up a mix of cgit and public-inbox, and adopt a quite different workflow.

Other projects I’m involved in, most importantly VoidLinux, are far more dependent on outside contributions and having access to CI infrastructure, which already makes it hard to move away from GitHub and Travis. For now we’ve decided to stick to GitHub, as there are more pressing issues currently, and we don’t expect GitHub to go haywire anytime soon. Still, our autonomy as an open source project is something we need to bethink more often and take care of.

NP: Julia Weldon—Comatose Hope

Copyright © 2004–2022