How did people even proof read their thousands of lines of code while typing it before AI LLMs? who the fuck has the patience for that?
typed languages and linters made everything fairly easyhard part was keeping all that complexity in your head when writing the next part of those thousands of lines, or figuring out exactly which line caused a segfault
>>109606369Did you forget a commit was between 1 and like 30 lines long you bitch ass Nigger
>>109606369You don't proofread code, you write bugs and then debug.It is actually harder with AI than it is without.
>>109606369Also, nobody was proof reading. Tech lead looked at the pull request to see if the coding style matches company guide lines and QA had to check if the shit works in most cases.Clueless jobless Fucknigger
You wrote the code and thus it's probably correct. And if code you wrote isn't probably correct, you're fired for being bad. This is basically how shops without QA worked forever.
>>109606369Without AI it's unsuprisingly much harder to write code while not knowing what the fuck you're doing.
>>109606369Does it compile? Does it do what it should? That's enough proof, no re-reading required.
>>109606369I used linter or shellproofs
>>109606410this is probably the best description of programming I've ever heard :D
>>109606369That's the neat thing, no one did. That's why they're malding now that AI is poking around their code and telling them it's shit.
>>109607995trvke
>>109606369Why do you think things like the bash bug and all the fucking CVE 999.99 linux vulns sat around in their open codebase for literally DECADES?Even the people writing the code can't spot vulns or backdoors in their own shit.
>>109606369times were different back then, frien
>>109606410If writing production code was still a part of my job, I would write the bugs myself and use AI to review my own code. These models are incredibly good at reviewing code even if they aren't all that good at writing it, and debugging my own code was always the worst part about the process.
>>109608041>These models are incredibly good at reviewing codeit feels like it would just gaslight me like "A-HA! You're using strcpy instead of strncpy! Bad boy"
>>109608109some of them are kind of presumptuous about what they consider an actual bug, but for the most part you can tune that away by making them prove runtime assertions and security impact.
>>109606369Nobody pushed 1000's of lines at a time back then.
Syntax errors you mean?
>>109606369For the love of god please just shut the fuck up and kill yourself.You have no idea how much of a favor you'd be the world if you and everyone as retarded as you would just kill yourself.Please do it for all our sakes. Buy a shotgun and blow your fucking head off.
>>109608109GPT does stuff like that a lot but usually you still only get a list of 10 or so items, so it's easy to spot the nits (Although it will call the nits blockers)
>>109606369>How did people even proof read their thousands of lines of code while typing it before AI LLMs?It was done very slowly and deadlines were multiple weeks instead of days or hours
>>109606369No one ever did which is why software has been on a non stop steady decline since the 80s.Occasional reminder there is not a single human on earth that has ever audited the Linux kernel post 2005.
>>109606369>editors with syntactic highlighting>proper testing during development irons out fuckups
