News for hackers. Stuff that matters.
Re:What git branching models actually work? by lmm on Tuesday July 30, 2013 @06:04AM (#6126075)
The problem I have with the CI approach is that it means you have to push a lot more of your program's integration to runtime, which is a worse place to detect failures. When adding a new feature ideally one writes the use case, which won't compile, and then implements the things it calls, which won't compile, and so on - if you're using types properly, then once you've added the front of a new feature your code won't compile until the entire feature implementation is correct. This is a wonderful way to code, eliminating huge classes of potential bugs, but CI seems fundamentally incompatible with it.
Re:What git branching models actually work? by calpaterson on Tuesday July 30, 2013 @07:47AM (#6126333)
Ok, so first of all, not everyone uses types and I for one think it is ok to use dynamically typed languages.

Second, I think you have a different vision of CI from me (maybe more frequent). I'm not advocating pushing code that is broken, I'm advocating pushing code that is working (but maybe is incomplete).