@tewodrosdy@SimonHoiberg gitops makes this a lot easier. Obviously it depends on your stack, but ideally more services just means copy + pasting more yaml.
@sorpaas@EIonAAusk To be totally honest, your attitude towards polkadot/JAM is so off-putting that I haven't even bothered to dig into JAR. Better implementation or not, show some respect to your inspiration and maybe you will receive the same!
@carllerche An even more extreme example of this is something like TypeScript. That can get even more chaotic, as the type system itself is insanely (dangerously) expressive.
@carllerche I don't have any particular examples, just anecdotal experience. My issue with expressivity and LLM's is the extra cognitive burden when reviewing as a human. In rust I can do 1 thing in many different ways/abstractions/etc. With golang, there is often 1 way.
@snowmaker being stuck was one of the most valuable times. It forced you to explore, and you would learn/uncover all sorts of things you otherwise would have not found. It was forced learning.
@oxcrowx Obviously this isn't always the case. But I remember going through the phase of development when I liked complex tools because I thought it gave me more power. Now I appreciate the benefits of enforced simplicity and constraint.
@oxcrowx There is a perception of creative freedom/power with the "complex" languages. Also ties nicely into the naive belief that complex = clever. I think a good analogy is the "all the gear - no idea" phenomenon you see IRL.
@ChShersh I think we're starting to observe the real cost. Forced to trust llm code, reduced awareness/understanding of codebases - we're handing over control. Dangerously, it also provides an illusive sense of alleviated responsibility. It's like a drug. Question is - how much do you care
@ibuildthecloud My previous workflow, which was to be more surgical and reference files with line-numbers, was probably more effective. I really want to believe the full "hands-off" approach is viable, but man I just seem to make more work for myself.
@ibuildthecloud I've just been asking myself the same question. I'm not convinced I'm any faster, at least for tasks where I know what the "correct" outcome should be. I spend so much time reminding AI what it should be doing.
@rockorager small surface area, not overly expressive, strong std library. I've found golang to be a pretty good candidate. Elixir also seems to do well.