Why a new VCS? There is no doubt that Git vastly improved our lives, and played a significant role in the advancement of software development over the past 18 years. But - it was built for a very different world in 2005 (slow networks, much smaller projects, no cloud), and is not the perfect tool for everyone today.
The biggest drawback of Git is its limited scalability - both in repository and file sizes, and the number of concurrent users. This is the reason Google and Meta built their own version control systems. It’s also the reason why other large companies, most notably in games development, semiconductors and financial services are still using legacy tools like SVN and Perforce.
Another issue we’re trying to fix is Git’s famous complexity. In our previous startup, a data scientist accidentally destroyed a month’s work of his team by using the wrong Git command (EDIT: we were eventually able to restore from a non-updated repo clone, after a few hours). As a developer who used CVS and SVN before Git was created, I often wondered why Git is so difficult to learn, compared to other tools.
On the other hand, Git’s branching and merging abilities are exceptional - this has enabled the modern software development methodologies that we all take for granted today (e.g. feature branches, CI/CD), greatly improving developers’ velocity.
We were wondering - is it possible to create an easy-to-use, fast, scalable version control system, with Git’s branching capabilities? And what else can be improved, while we’re at it?
One thing available in modern cloud tools is real-time collaboration (e.g. Google Docs, Figma). While developers don’t necessarily want their work in progress to be visible to everyone, it may be very useful to easily share it when you want to get feedback before a commit, to detect and prevent merge conflicts, and to have visibility into which parts of the codebase are being changed by others.
Diversion is built on top of distributed storage and databases, accessible via REST API, and runs on serverless cloud infrastructure. Every repository operation is an API call (commit, branch, merge etc.). The desktop client synchronizes all work in progress to the cloud in real time (even before a commit). Users can work with Diversion using an interactive CLI, Web UI, or IDE plugins (currently JetBrains, more coming soon). The Web UI allows to perform most basic operations, without needing to install a desktop client.
Diversion is compatible with Git, and can synchronize with existing Git repositories (each new commit in Diversion goes into Git, and vice versa). We’re planning to release it as open source once the code base matures, and when we implement an open source repositories directory on our website (naturally, Diversion’s code is managed on Diversion!)
We’re in open beta, you can try it here (https://diversion.dev) (click Get Started). It’s completely self-service and there’s no need to talk to anyone, and it’s free for small teams (https://diversion.dev/pricing).
Building a version control is hard (as we have learned), and Diversion still has a long way to go. We are currently working on improving speed, CI integrations, plugins to IDEs and game engines, and other usability improvements. We would love to hear your thoughts and feedback on what we’ve got so far!
by Andrey
Azimov