>>105864849
>You don't choose language "because AI is good with it".
For new projects, sure you do. It's going to be a serious consideration from now on.
Typescript is actually a decent language, subject to some caveats (need to use good tools, specifically a very strict eslint config). Having used both, I'd definitely rather write TS than C#.
>Legacy codebases aren't a thing that you can simply "just refactor".
You can. It involves a lot of work and some risk, but it's perfectly possible. It just depends on where your priorities lie.
If you're stuck with an ancient pile of shit that you can't use modern tools on and which AI can't understand, and your competitors have a cleaner codebase which they can change more easily and with more confidence, then you're going to fall behind.
>microservices
I actually prefer monolithic if it's structured well, and when microservices are used I prefer when they're stateless.
But identifying a chunk of functionality in your shitty codebase to replace with calls to a new, well-written microservice can be a decent way to make some kind of progress.

Back | Go to board

Board:

/g/

Thread:

Randomized controlled trial finds developers are 19% slower if they use AI