AI is not perfect there is some bad review comments. As a PR author you can reject such comments by resolving them.
My idea going forward is that I will not review a PR manually if there are open AI reviews.
I want to limit the amount of manual reviews because;
for reviewers it takes a lot of time. I don't want to waste time on nits that we could avoid automatically. I have to re-review the whole PR every time I see some nits.
for developers manual reviews are slow it is better to get feedback directly.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
To me this talk has lived rent free in my brain since 2012 https://youtu.be/5wwKjdvPxa8 its about how CI can be used as a code reviewer. Most CI systems are pass fail. But there tea leaves that often go unnoticed. I think in the age of AI its probably a good idea to use AI as a first pass aproach to keep from getting overwhelmed. I even know of an open source project that is talking internally about poising their skill file to say all issues must be written using Shakespearen English. That way they can prioritize issues written by people over issues written by AI. All that being said I would see what you can automate information wise to help speed up code reviews. An example would be call graphs. Next is what unit test covers this function change. Those are two things that are very useful to know while reviewing code that often breaks my train of thought when I have to hunt them down while reviewing code. Also here is a post I made along time ago about how I abused the cppcheck misra plugin for code reviews of code that doesn't have to be misra complaint, but I still find it useful for my brain to think about https://sourceforge.net/p/cppcheck/discussion/development/thread/834110f0e7/?page=2&limit=25#aacc
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I am adding AI reviews in opensource cppcheck
AI is not perfect there is some bad review comments. As a PR author you can reject such comments by resolving them.
My idea going forward is that I will not review a PR manually if there are open AI reviews.
I want to limit the amount of manual reviews because;
To me this talk has lived rent free in my brain since 2012 https://youtu.be/5wwKjdvPxa8 its about how CI can be used as a code reviewer. Most CI systems are pass fail. But there tea leaves that often go unnoticed. I think in the age of AI its probably a good idea to use AI as a first pass aproach to keep from getting overwhelmed. I even know of an open source project that is talking internally about poising their skill file to say all issues must be written using Shakespearen English. That way they can prioritize issues written by people over issues written by AI. All that being said I would see what you can automate information wise to help speed up code reviews. An example would be call graphs. Next is what unit test covers this function change. Those are two things that are very useful to know while reviewing code that often breaks my train of thought when I have to hunt them down while reviewing code. Also here is a post I made along time ago about how I abused the cppcheck misra plugin for code reviews of code that doesn't have to be misra complaint, but I still find it useful for my brain to think about https://sourceforge.net/p/cppcheck/discussion/development/thread/834110f0e7/?page=2&limit=25#aacc