How Has C++ Good Practice Advice Changed in a World of AI Code Agents
I'm one of a group of industry veterans who have been pumping out "good clean C++ code advice" talks at ADC (and elsewhere) for many years.
And amongst those of us who preach about this kind of thing, there's always been a solid consensus on what "good practice" looks like for C++ and coding in general.
But then code agents popped up and surprised everyone by doing things none of us would have believed possible. Should we now be reframing any of the advice that was previously considered to be a no-brainer?
If you're using a code agent to work on an existing codebase, how much is it worth changing your style to get the best results?
If you're embarking on a new project, and expect to use AI in writing it, what considerations should you think about from the very start?
I'll cover things like:
- do we need to change any assumptions about code techniques?
- are there things that we used to worry about that no longer matter?
- how do we safely let agents loose on a codebase
- how to use these things to improve your own style and quality
- what are some things that you might want to do now, that you couldn't be bothered to before
As well as my own thoughts, I'll be drawing on input from various industry gurus. And I'll be referencing examples from work that I've been involved in on both old and new codebases.
Julian Storer
Founder
Cmajor Software Ltd
I'm the creator of JUCE, Tracktion, Cmajor and CHOC 🙂