2028: No code written using more than one monitor. Excessive screen real estate reduces human oversight.
2029: No code written in editors with a light theme. Insufficient eye strain suggests inadequate review.
I don't really see the appeal besides free hosting + some storage.
People should either self-host a gitea/forgejo/gogs server for like 3$ a month on hetzner or just use gitea.com.
If you want to share your repos, just use sharemygit.com and spread the repo on coding forums. Personally I find this a much better stimulant for a community, then just having free-flowing code spread around like dust under a bed.
+ you actually dont need to be the victim of the whims of some ideological git-server provider. Ahem github (master->main), codeberg (no private repos are allowed), gitlab (you have to pay) etc. etc.
jaggs•2h ago
If so, it seems kind of short sighted. Within a very short period of time all code will be AI code. What then?
And as the models improve, along with better code will come better bug fixing etc etc. So the quality of AI code will absolutely surpass that of human code. What then?
Will the repositories demand proof of programming ability? Why, when AI will be handling everything anyway. It would be rather like driving schools mandating that all drivers can strip an engine before they're allowed to drive? Very odd.
Hackbraten•51m ago
InsideOutSanta•42m ago
If they change their opinions in the future, they can reevaluate their stance and update their terms.
I love using LLMs for coding, but I think it's absolutely fine if people make different decisions. In fact, I think trying to maintain code repositories that are free of LLM-generated code is extremely valuable.
mschuster91•29m ago
It took decades of manufacturers ramping up reliability to the point that driving schools didn't teach you how to change spark plugs, oil swaps and how to change a tire on the road side. I did my license around 2010 in Germany and changing a tire plus how to check for oil and coolant levels was part of the curriculum.
And similarly, AI has to go through that "ripening" as well before it can be trusted. At the moment, there is no solution even found on how to prevent hallucinations or in-band signalling attacks, much less actually validated at scale.
vrganj•26m ago
This is unfounded speculation, presented as inevitable fact.
inigyou•21m ago
onesandofgrain•6m ago