`git bisect run ...`
Has been really nice for me when trying to find which which commit broke long running commands. It'll run atomically using the return code for good/bad determination.
I remember learning about this a few years ago, and it felt like both a very obvious feature to have and a superpower at the same time.
Just a few easy steps between "I can test for this bug and don't know where it is in hundreds of thousands of lines" and "Here's the three line change that created the bug and context about them".
`git bisect run ...` Has been really nice for me when trying to find which which commit broke long running commands. It'll run atomically using the return code for good/bad determination.
It's all good and fun until your build command takes 30 minutes to run and you still need manual testing :(
I remember learning about this a few years ago, and it felt like both a very obvious feature to have and a superpower at the same time.
Just a few easy steps between "I can test for this bug and don't know where it is in hundreds of thousands of lines" and "Here's the three line change that created the bug and context about them".
Pretty cool. Knew this existed but not really how it is a solution.