Introducing PR Alert Scans
With Rover alert scans, catch performance and deployment risks before they get to prod.
Your team wants to ship fast. More PRs, shorter cycle times, more deploys. But code churn impacts reliability. More bugs, shorter user retention, more request latency.
At Rover, we’ve been thinking hard about why reliability when code throughput grows (either through productivity or a bigger team), and how to solve it.
Simply put, each engineer’s understanding of the code gets worse over time.
As the team grows, each individual sees the codebase through a smaller window, focusing only on that part. As development speed increases, what your engineers knew about the code 2 weeks ago no longer holds up.
Our limited, personal context windows are not helped by the standard code review process, either. Each PR only shows what’s changed - usually a good thing, to reduce noise. But with increasingly complicated development environments, a simple change might get a “LGTM”, but ends up breaking prod due to some funky interaction with a completely different service.
Introducing PR scans with Rover
Starting today, Rover will automatically scan every pull request in the context of your entire system and flag any risks it will have on the code as a whole.
Your team will be alerted to:
Performance bottlenecks across code chains
Bugs occurring at service boundaries (e.g. APIs; database tables)
Mismatches between what the “working” code is doing, and what the intended business logic is
…and much more
This all happens in your PR so you don’t have to convince your engineers to switch up their workflow - we’ll come to you.
How it works
Under the hood, Rover continuously indexes your codebase into a graph of APIs, database usage, service calls, and infrastructure references.
When a PR is opened, we analyse touched code paths for service impacts, and run multiple agents to think through potential consequences of the code diff.
It’s that simple! (it’s not really, but that’s our problem to deal with - you can sit back and focus on shipping)
What’s next
Currently, Rover PR scans catch a wide variety of issues across major programming languages. But each codebase is unique, and needs its own special line of inquiry from time to time.
Learning about your codebase. Rover is learning to learn! When you connect a codebase, Rover will start to build up a picture of the particular quirks of your codebase so it can identify more relevant issues. So if, like us, you happen to name your microservices after Tolkein characters, Rover will know that your PR mentioning mithrandir isn’t some Silmarillion fanfic…
Bespoke alert scans. Need to hone in on reliability issues ahead of your next major prod deploy? Or diagnose some bugs you’ve been having with your signup flow? We’re working on bringing Rover’s scanning capabilities outside of the PR, so you can get the exact assistance you need, when you need it.
Dive into the code graph. Rover’s alert scanning capabilities is built on top of a graph of your codebase, which it constructs on every push. This graph will soon be made available for you to explore.
Want to see how Rover can help your team improve code reliability?
Try it now
Rover PR alert scanning is live for all users, for free.
It takes just a minute to connect your codebase, and then Rover will start detecting code risks at your very next PR - no further config is required.
If your team isn’t using Rover yet (why not?)….
👉 Get your team setup in 2 clicks
Great stuff!