I see a future in jj
In December of 2012, I was home for Christmas, reading Hacker News. And that’s when I saw “Rust 0.5 released."" I’m a big fan of programming languages, so I decided to check it out. At the time, I was working on Ruby and Rails, but in college, I had wanted to focus on compilers, and my friends were all very much into systems stuff. So I decided to give Rust a try.
And I liked it! But, for other reasons I won’t get into here, I was thinking about a lot of things in that moment. I was looking to shake things up a bit. So I asked myself: is Rust going to be A Thing?
Why I chose Rust
So, I thought about it. What does a programming language need to be successful? It needs some sort of market fit. It needs to have people willing to work on it, as bringing a new language into the world is a lot of work. And it needs users.
When I considered all of these things, here’s what I saw with Rust:
Market fit: there was basically no credible alternatives to C and C++. I had been involved in the D community a bit, but it was clear that it wasn’t going to take off. Go was a few years old, and hit 1.0 earlier that year, but for the kinds of work that C and C++ are uniquely able to do, I saw the same problem that I did with D: garbage collection. This doesn’t mean Go isn’t a good language, or that it’s not popular, but I didn’t see it as being able to credibly challenge C and C++ in their strongholds. Rust, on the other hand, had a novel approach to these problems: memory safety without garbage collection. Now, I also need to mention that Rust back in those days was much closer to Go than it even is today, but again, I had just learned about it for a few hours, I didn’t really have a deep understanding of it yet. If I had, I actually might have also dismissed it as well, as it wasn’t really GC that was the issue, but a significant runtime. But again: I hadn’t really come to that understanding yet. Point is: low-level programming was a space where there hadn’t been much innovation in a very long time, and I thought that meant that Rust had a chance. Check.
For a team: well, Mozilla was backing it. This is a big deal. It meant that there were folks whose job it was to work on the language. There’s so much that you need to do to make a new language, and that means a ton of work, which means that if you’re going to be able to get it done in a reasonable amount of time, having paid folks working on it is certainly better than the alternative. Check.
And finally, how does this translate into users? Well, Mozilla was planning on using it in Firefox. This is huge. Firefox is a major project, and if they could manage to use Rust in it, that would prove that Rust was capable of doing real work. And, more importantly, it would mean that there would be a lot of folks who would need to learn Rust to work on Firefox. This would create a base of users, which would help the language grow. Check.
Finally, even though it wasn’t part of my initial assessment, I just really liked the Rust folks. I had joined IRC and chatted with people, and unlike many IRC rooms, they were actually really nice. I wanted to be around them more. And if I did, other people probably would too. So that was also a plus.
So, I started learning Rust. I decided to write a tutorial for it, “Rust for Rubyists,” because I’m a sucker for alliteration. And I eventually joined the team, co-authored The Book, and if you’re reading this post, you probably know the rest of the story.
Enter jj
For some background, jj is a new version control system (VCS), not a programming language. It is written in Rust though! While I talked about how I decided to get involved with Rust above, my approach here generalizes to other kinds of software projects, not just programming languages.
I have a rule of thumb: if Rain likes something, I will probably like that thing, as we have similar technical tastes. So when I heard her talk about jj, I put that on my list of things to spend some time with at some point. I was especially intrigued because Rain had worked at Meta on their source control team. So if she’s recommending something related to source control, that’s a huge green flag.
It took me a while, but one Saturday morning, I woke up a bit early, and thought to myself, “I have nothing to do today. Let’s take a look at jj.” So I did. You’ll note that link goes to a commit starting a book about jj. Since it worked for me with Rust, it probably would work for me for jj as well. Writing about something really helps clarify my thinking about it, and what better time to write something for a beginner than when you’re also a beginner?
Anyway, people seem to really like my tutorial, and I’m thankful for that.
The future of jj
So, what do I see in jj? Well, a lot of it kind of eerily mirrors what I saw in Rust: a good market fit, a solid team, and a potential user base.
But the market fit is interesting. Git has clearly won, it has all of the mindshare, but since you can use jj to work on Git repositories, it can be adopted incrementally. At Oxide, Rain started using jj, and more of us did, and now we’ve got a chat channel dedicated to it. This is, in my opinion, the only viable way to introduce a new VCS: it has to be able to be partially adopted.
Google is using jj, and so that is a bit different than Mozilla, but the same basic idea. I have more to say about Google’s relationship to jj, but that’s going to be a follow-up blog post. What I will say in this post is that at the first ever jj conference a few weeks ago, Martin (the creator of jj) said that internal adoption is going really well. I’m burying the lede a bit here, because the video isn’t up yet, and I don’t want to get the details of some of the more exciting news incorrect in this post. I also don’t mean to imply that everyone at Google is using jj, but the contingent feels significant to me, given how hard it is to introduce a new VCS inside a company of that size. Well, in this case, it’s using Piper as the backend, so you could argue about some of the details here, but the point is: jj is being used in projects as small as individual developers and as large as one of the largest monorepos in the world. That’s a big deal. It can show the social proof needed for others to give jj a chance.
Outside of Google, a lot of people say that there’s a bit of a learning curve, but once you get over that, people really like it. Sound familiar? I think jj is different from Rust in this regard in that it’s also very easy to learn if you aren’t someone who really knows a ton about Git. It’s folks that really know Git internals and have put time and care into their workflows that can struggle a bit with jj, because jj is different. But for people who just want to get work done, jj is really easy to pick up. And when people do, they often tend to like it. jj has developed a bit of a reputation for having a passionate fanbase. People are adopting it in a skunkworks way. This is a great sign for a new tool.
And finally, the team. Martin is very dedicated to jj, and has been working on it for a long time. There’s also a small group of folks working on it with him. It recently moved out from his personal GitHub account to its own organization, and has started a more formal governance. The team is full of people who have a deep history of working on source control tools, and they know what they’re doing. The burgeoning jj community reminds me of that early Rust community: a bunch of nice folks who are excited about something and eager to help it grow.
Basically, to me, jj’s future looks very bright. It reminds me of Rust in all of the best ways.
Putting my money where my mouth is
Speaking of burying the lede… I’ve decided to leave Oxide. Oxide is the best job I’ve ever had, and I love the people I work with. I was employee 17. I think the business will do fantastic in the future, and honestly it’s a bit of a strange time to decide to leave, since things are going so well. But at the same time, some of my friends have started a new company, ERSC, which is going to be building a new platform for developer collaboration on top of jj. Don’t worry, “errssk” isn’t going to be the name of the product. It’s kind of like how GitHub was incorporated as Logical Awesome, but nobody calls it that.
This won’t be happening until next month, I have some stuff to wrap up at Oxide, and I’m going to take a week off before starting. But as sad as I am to be leaving Oxide, I’m also really excited to be able to spend more time working in the jj community, and helping build out this new platform. For those of you who’ve been asking me to finish my tutorial, well, now I’ll have the time to actually do that! I’m sorry it’s taken so long! You’ll see me talking about jj even more, spending even more time in the Discord, and generally being more involved in the community. And I’ll be writing more posts about it here as well, of course.
I’m really excited about this next chapter. 2025 has been a very good year for me so far, for a number of reasons, and I am grateful to be able to take a chance on something that I’m truly passionate about.
Here’s my post about this post on BlueSky:
I see a future in #jj-vcs: steveklabnik.com/writing/i-se...