Gregory's Server
|
22 posts total
12
Its honestly ridiculous that so many people seem to think that if you maintain libraries in rust or clojure or typescript its "production software", regardless of whether its deployed in production, but if you maintain a library with the same functionality in any lisp thats not clojure its "a hobby", even if its an emacs package used every day by countless programmers. but then again, most production software is bad, and most things people put their heart and soul into tends to be good (: what corporations demand shouldn't be the metric of software quality. developer bitch and groan about mainstream software practices and then hate on anything that actually goes against the grain.  @rml the whole way of thinking and talking where there are reified concepts of “senior software engineers” and “best practices”, free-floating ideas of “beginner friendly” and “production quality”/“production ready”, along with a whole class of ever-changing meaningless truisms “actually abstractions are bad”, “actually comments are bad” “everything is a trade off”makes me despair - people are genuinely convinced there is something meaningful about all of the above and become very angry if you doubt the framing
Show previous comments
  @rml You know when the religious types get all hyper and start speaking in tongues and dancing around at the magical touch of Jesus? That's how any heavy-duty Haskell/ML stuff sounds to me. if you think the former is more desirable than the latter, then I got a theorem prover to sell to you dressed as a programming language  @rml I was a huge fan of #haskell & type theory. That's how I got into "advanced programming languages". I tinkered with it over a year and end up considering it totally useless. In (kinda&-)functional programming we have just functions and data. MLs praises functions. Lisps praises data. And I prefer data. Boss: "can we add #ChatGPT there to speed up production?"  People are seeing better results in less time while having more fun targeting the latter. But if its not a massive industrial pile of fast moving complexity seeped in corporate interests it cant be very serious, right? I mean, 4mb? C'mon, thats not a very serious compiler. I want my compiler toolchain to be big, strong, manly.  @rml as much as I don't care for Common Lisp, SBCL is a solid piece of engineering work @daviwil never heard of this, I have a big stack of books lined up rn so it will be a while before I get to it, but you should do a series of streams working through some of the ideas! This #conj23 talk is great, Sam Richie going through his journey diving into Sussman's #SICM and #scmutils, it will be very relatable to anyone who has become possessed for several years with the drive to truly understand some body of knowlege (I'm looking at all of you) When I first go into #guix, I experimented with every little thing. Now I only change my configuration every three months or so, while for anything else I just use guix shell. And my system remains solid and predictable, with no weird state and quick environment variable tweeks building up leading to library collisions etc that can come to haunt you at a moments notice. Just my studio, as I designed it to be. This has meant that I havent contributed much in a while. But I'm honestly not the type to try out the newest software as its released. Everything I need is packaged in guix already. And of all the things I've packaged, I never started actually using any of them lol. So I'm kind of glad I didn't submit them upstream, because im not sure if we really need to include any packages that noone is using.  How do you manage your Emacs packages? Do you have a core set under your profile and then load others on demand via Guix shell? Or do you have a manifest that you use for each project as needed? I declared Emacs bankruptcy and am considering moving to Guix and starting from scratch. Suggestions welcome #guile tip of the day: use module-map to create a list of all the functions in a given module #lurk, a #scheme-like content-addressed #lisp for "Zero Knowlege Proofs" 12
|
@rml If anyone else ever realized C++ was complicated, we wouldn’t need him to save us.