Can a flavor of Lisp become the new defacto shell for Linux?
>>109172193no, too many ((())) are a pain to write. but you can try schemesh and see for yourself.
>>109172193Technically yes, anything is possible in software, but also no
>>109172193Linux? NoGNU+Linux? Yes
>>109172193yeah surehttps://scsh.net/
>>109172493https://doc.guix.gnu.org/guile/latest/en/html_node/The-Scheme-shell-_0028scsh_0029.html
>>109172193how about the C64 "shell", basic, peeks and pokes, sys(32760) or someshit. gotta put in the work to get the gains
Software idea: replacing bash with BBC BASIC
>>109172193eshell is pretty interesting, but it only runs inside of emacs.
>>109172193I think LISP is going to eventually have a major resurgence as AI continues to progress.There's a reason LISP machines were the original AI/ML machines back in the day. The structure of LISP is more "natural" to the way machines want to think.So, right now LLMs are really good at giving us stuff like Python because theres a ton of human-generated training data for it, and because the LLM knows Python is very human-readable. But the LLM's doesn't really want to write Python. Python sucks from a computer's perspective.As time goes on and AI-powered coding becomes the norm, LLMs are increasingly going to use LISP-like syntaxes that are hard for humans to read but are incredibly machine-efficient and token-dense. It will happen subtly at first, but there'll be a tipping point and pretty quickly nearly 100% of code will be written in this universal LLM-LISP language.
>>109172193We have an OS with a Lisp shell, it's called Emacs
>>109173118Considering this it's kinda funny how bad LLMs are with Lisp, can't do ((())) properly
>>109173174Yeah, and I think that's just because there is not enough LISP out there. Like, only 0.1% of github code is LISP. So, there needs to be more for LLMs to train on. But once that happens I think the shift will happen.And LISP programmers are using LLMs to produce greater amounts of working code than they could before. So I think we are going to hit a point where the training data becomes sufficient and LLMs get as good at one-shotting LISP as they currently are at python/javascript. And then, suddenly, the LISP code will start looking *better* than the python code....
>>109172193It call emacs and it sucks dick.As much as sh derivatives are bad, almost no one wants to program or script in lisp. Even lisp weenies spend there time admiring lisp's simplicity more than program in it.>>109173118>I think LISP is going to eventually have a major resurgence as AI continues to progress.>There's a reason LISP machines were the original AI/ML machines back in the day. The structure of LISP is more "natural" to the way machines want to think.How computers "think" is doing a lot of math fast and following sequential instuctions.The type of AI running on LISP machines has little to do with ML or neural nets and would have been rules engines and symbolic processing.
>>109173267following sequential instructions.... sequential instructions like... a list? Would a list processing language be appropriate for that, do you think?
Closest we have atm is gash, but that is only useful for bootstrapping GuixAs a daily driver shell? Wouldnt rule it out some day
>>109173118Nah, you are delusional. Large Language Models are trained. They are token predictor machines, not intelligent. There is no pure emergent intelligence or learning there no matter how much some ((corporation)) would like you to think otherwise.
>>109173299following sequential instructions.... sequential instructions like... a procedure? Would a procedural language be appropriate for that, do you think?This is why I can't take lispfags seriously; for every nice bit of tooling they invent or heterodox perspective on a problem, they attribute everything to the properties of language itself.Brendan Eich is the most intelligent lisper on the planet because he realized the semantics people cooked up are nice, but the language is ultimately irrelevant.
>>109173395Fuck, 4chan ate my meme arrow
>>109172193If you got some way to neatly and quickly convert most bash scripts then maybe if there's market acceptance
impossible my nigga, impossible. see, the syntax, hahahaha.though i see, you see the idea, but it should be a system-wide runtime.
>>109172193all those brackets make it look antisemetic
>>109172193Truth bomb for ya. Ready? Sit down.No one uses Lisp. Several people here on 4chan loudly proclaim that they use Lisp and how great it is but they really don't do anything beside hello world programs. Sometimes someone copies a fizzbuzz program that someone else did and uses it to proclaim how they are so proficient in Lisp, but that's a sham.Back in the late 70s Stallman claimed to use it and a few academician's looked at it in a clinical way for its unusual structure, but have since moved on to better things.So why would one want to replace bash with an obscure language that is bad and that no one actually uses?
>>109172193It's nice. I've been doing it for years.
>>109172193No. Lisp is not a good shell language. Shell languages need to be ergonomic for common interactive OS tasks.
>>109175278What shell is that?
>>109175089none of that is true. in fact the only people who don't use lisp are people who lack individuality, chase trends and make no significant contributions to prevalent programming standards. your notion of "bad" and "no one uses it" is just npc programming.
>>109176854You're both wrong.Lisp is good and important but you fags are seriously delusional about the relevance and value of other languages.
