Join devRant
Do all the things like
++ or -- rants, post your own rants, comment on others' rants and build your customized dev avatar
Sign Up
Pipeless API
From the creators of devRant, Pipeless lets you power real-time personalized recommendations and activity feeds using a simple API
Learn More
-
yeah the auto complete is nice
also helps when refactoring / rewriting. it just shows me the old code so I don't have to look for it
AI is just a better search I swear. especially in this era of garbage SEO and low quality data it's actually just useful because the people who made it put in the effort to find higher quality data. literally its use is only because it fixed a problem malicious humans made out of selfishness. sigh. I miss early 2000s -
@jestdotty it is better in some areas, but is _actively_ worse in others. ChatGPT writes good python code, but if you ask it to fix something in linux, it will most definitely offer you some hacky things that outright don't work, while linux forums would've offered you good advice
GPT-4.5 is suspiciously good at fixing linux though. it's a shame that it's so expensive to use -
@kiki opus is god
all Linux is hacky and doesn't work like 80% of the time. or it does work but it's hacky so it doesn't work exactly like you wanted it to work. just sounds normal to me
my Linux knowledge is like my knowledge of CSS. I've exclusively used both for years but I know nothing about them 🥁
Tried Cursor. It's "Agent" thing have given me the whole new appreciation for my existing codebase. For how every line of code and every micro-decision I make is deliberate. Though it did it in the worst way possible — by "optimizing" my code that I didn't request to be optimized.
I feel like I delegated a surgeon's work to a lawnmower roomba. Yes, both of them do cut stuff, but are they interchangeable?
Cursor's autocomplete is stellar though. Because of how consistent my code is, it almost never fails. It baffles me that the codebase that is consistent enough for the AI autocomplete to never fail somehow causes the same AI, but acting as an agent instead, to completely shit the bed.
My code is consistent because I copy and paste a lot. But, because of how expressive it is (thanks to my zero-framework approach), I only ever need to copy and paste ten lines or so max to do what I need. When I say that I "copy and paste code", what comes to mind? I bet it's acres of boilerplate. Not here.
I'm now applying Orwellian newspeak rules to my naming. For a distinct entity, I create the shortest possible name and try to carry it everywhere I go. Yes, short names will run out, but naming entities differently to avoid conflicts is the job for future me. Premature optimization is the root of all evil, and thinking about variable names in advance is premature optimization too.
For example, if the user entity in the database has lastName field, its variable in imperative code will be the same, and the form input name will be lastName too. Why use "userLastName" and "lastNameInput" or "lastNameField", or even "userLastNameField"? YAGNI.
I'm inching closer and closer to a universal gray goo architecture that can absorb anything. Wait till I replace what I copy and paste with brand new keywords and create a new language. I've already started doing that with the way I write util functions.
random