Paul Snively
Paul Snively
33.6K posts
Paul Snively
@JustDeezGuy
Just deez guy, you know?
Earth, for nowJoined March 2024
Paul Snively’s posts
Replying to
“I have a type and I like it, but I married the best one” is kinda a W, IMO.
Replying to
You know you’re dealing with a different kind of action movie when the protagonist’s first words under stress are “THINK, John! THINK!”
Replying to
Besides that, "a gay man's satire on masculinity" is literally "Fight Club."
Replying to
If you watch the documentaries for “The Incredibles,” you learn Brad Bird was inspired by overhearing neighbor ladies insult his stay-at-home wife and mother of their children, when it was clear to him she was a hero. A backyard BBQ version of this got as far as storyboards, and
"We do CI/CD with GitHub Actions:" insta-red-flag.
Quote
Rick the Dad
@rickasaurus
this is my first experience with github actions and it's like they say: sometimes free is the most expensive option
Let’s also reflect on the fact the Firefox team were very experienced C++ developers, had the ability to hire the best, had a 250KLOC+ C++ CSS engine they tried to make concurrent because CSS benefits a LOT from it, suffered from massive memory-management issues, and decided it
Quote
void__zero2
@void_zero2
there is a surprisingly large amount of people that think not being able to write memory safe C/C++ is a "skill issue"
I want all those people to watch & listen to this video of Azure CTO very carefully
Google, Microsoft and others are very open that C/C++ is a problem
0:36
Replying to
Yes.
Also because it’s easier to PRETEND to eat from an opaque container. Virtually all scenes involving eating are simulated, to avoid full mouths, digestive issues later, etc.
Haskell doesn't make I/O hard; it reveals that I/O is hard.
Rust doesn't make manual memory management hard; it reveals that manual memory management is hard. x.com/mzndr_dev/stat
You’re unable to view this Post because this account owner limits who can view their Posts. Learn more
Replying to
*checks notes* Academy Award winner Mira Sorvino has entered the chat.
Haskell doesn't make I/O difficult; it reveals that I/O is difficult.
Rust doesn't make manual memory management difficult; it reveals that manual memory management is difficult.
Quote
Computer Science
@CompSciFact
One time Dijkstra was accused of making programming more difficult and responded that he was only revealing how difficult programming is.
x.com/AlgebraFact/st
Replying to
ngl, I love that she stops herself from saying "I'm just fucking with you."
Related:
"Haskell doesn't make I/O complex. It reveals that I/O is complex."
Quote
Julia REPL stan now at mas to/b sky
@miguelraz_
Rust surfacing hidden complexity and making it explicit is part of its superpowers x.com/0xglitchbyte/s…
Here’s what I find frustrating:
Early in computer science as a discipline, it was understood that we needed to be able to reason about code. People like Dijkstra and Hoare described how to do it. One of the results was Floyd-Hoare Logic, which supports preconditions and post x.com/krismicinski/s
You’re unable to view this Post because this account owner limits who can view their Posts. Learn more
Replying to
Multi-player games aren't anti-social, & have led to lasting IRL friendships & even marriages.
Consumption-based: so are movies, which also remain a popular date-night activity.
Replying to
"Breaking Bad" is pure, unadulterated Greek tragedy. For Walter to be sympathetic—at all—you have to have no exceptions to his hubris leading to his downfall. He HAS a good life with Skyler & junior, & COULD keep it or even return to it & redeem himself. His hubris prevents it.
Hot take: there's exactly zero reason for someone "halfway through their CS degree" to know "you could pass structs directly to functions in C," since that fact has literally nothing to do with computer science.
Quote
east coast anna
@microsoft_worm
just interacted with someone halfway through their CS degree at a top 20 school who didn’t know you could pass structs directly to functions in C. however bad you think it is, it’s worse
Replying to
I teared up when deputy Andy fell apart trying to take pictures, and it occurred to me he was really the only one acting appropriately under the circumstances.
And there's a lot of that in "Twin Peaks." Even when, at first, Leland's reactions seem mentally unhinged, we later
This is what people mean when they say learning Haskell, OCaml, Scala… made them a better programmer in more popular languages.
Once you HAVE TO think about your code as nothing but transformations on immutable data, and have a strong intuition as to what “functional core,
Quote
Jules
@analytichegel
I'm a big believer that even in languages like C you should architect your codebase like you're writing Haskell for exactly this reason. x.com/nicbarkeragain…
In particular, Firefox is increasingly in Rust, and Unreal Engine 6 will increasingly be in Verse.
It’s certainly not indicative to refer to 20-30 year old projects and say “See!? C++!”
Quote
HSVSphere
@HSVSphere
These people would have a heart attack if they knew what the creators of these pieces of software are trying to do to C++ and what language they're moving to.
C++ is outdated, deprecated technology. What you're listing here is tech debt. x.com/tiwaryash/stat…
Unpopular opinion: use Protocol Buffers, not because the format is that great (it isn't), but because the range of supported languages for codegen can't be beat.
Further unpopular opinion: Protocol Buffers plus named pipes (Windows) or UNIX-domain sockets (everything else) are a
Quote
gingerBill
@TheGingerBill
Replying to @SebAaltonen
The people who use JSON outside of JS probably don't know that you can just send a binary blob across a network.
I had to explain this to someone the other day, and they were actually surprised that it was that simple.
Hi folks!
It is with no small amount of sadness that I report that a contract-to-hire opportunity I was very excited about has fallen through for what sound like the simple, but brutal, economic reasons faced by all businesses, but especially very young startups still striving
Bird was clearly... maybe not "angry," exactly, but upset in some way. The storyboards in the documentary reflect exactly what you're describing: the neighbor characters are outright harpies. It's good that scene didn't make it in—it was far too bitter and on-the-nose for an
Replying to
Nevermind the writing. I just love that Ocean literally doesn't care about the money and only goes through the whole thing to prove to Tess Terry doesn't love her.
Replying to
Non-existent fit for purpose (you end up writing a dependency/task graph by hand, with a bespoke assembly process for each artifact, in essence). No real support for incremental building. No real support for caching. No real support for observability, especially in failure cases.
I like Rust. Really. I do.
But the list of things they compromised on (to be fair, because, uh, we don't know how to do them all at the same time) in order to have an affine type system, on top of which they built the borrow checker, is... pretty long. And higher-kinded types
Quote
ULTRA/MAGIC
@ULTRAMAGlC
If rust had hkt I'd use it a lot more
I don't want to hear another damned word about how complex and/or unnecessary monads are.
Quote
Graham Christensen
@grhmc
GitHub Actions has INSANE semantics: a "Required" job is considered to have succeeded if it is skipped. Jobs are skipped if they `needs` a job that failed.
A major reason I couldn't take Go seriously: to design a language without parametric polymorphism AFTER both C++ and Java had to bolt it on after the fact, excruciatingly, with weird corner cases Standard ML, OCaml, Haskell, Scala... don't have was just too aggressively ignorant.
Quote
спедру
@spedru
the go team literally had to bring wadler on board to fix their dogshit type system for 1.18 and i say this as a go defender. come on man x.com/kai_fall/statu…
Replying to
I mean, yes. But then explain the Linux kernel anti-Rust hysteria. (To be fair, Linus himself seems to have very moderate, reasoned thoughts about it.)
Quote
Jonathan Blow
@Jonathan_Blow
Replying to @effectfully
If a lot of your functions can fail, you are a bad programmer. Sorry.
I use Haskell on the job, & far & away the most frustrating thing about it is the (small) army of people who have clearly never used it & don't understand it, but feel entirely confident in saying things like:
* No one understands monads
* The IO monad makes your code impure
Quote
Fay λ
@fay_carsons
The misinformation about Haskell is wild, stuff like “you can’t print debug” and the attitude that it’s only for white-papers.
I’ve been using it pretty much exclusively in my personal projects recently and not only is all of that untrue, it is literally so fun to write??
Your periodic reminder Haskell is a RESEARCH LANGUAGE.
If you use it professionally (I do!) one of your primary responsibilities is to curate your library choices, lock down your compiler version, identify a fixed set of extensions not open to further debate…
In other words,
Quote
Dmitrii Kovanikov
@ChShersh
Replying to @yoobinray
The main Haskell problem - it tries to do too much.
Type level programming, fast low level optimisations, innovative build tooling, 5 code formatters, constantly changing the standard library, bespoke huge CI infrastructure just for GHC, novel GC algorithms, runtime system,
Show moreThe hatred every programmer I know has for Jira would, all by itself, prevent me from adopting it if I were a technical cofounder.
Quote
Nic Barker
@nicbarkeragain
I think it's underappreciated how much more productive programmers are when they're having fun. It sounds ridiculous but I think in general when businesses add extra process, admin, rules etc that make the work less fun, they fail to appreciate that it will impact productivity.
Some programming paradigms, such as functional programming, are objectively superior to others.
Quote
Slim Jimmy
@slimjimmy_dev
what do you believe, deep in your bones, about programming that almost everybody else does not believe?
The problem is the type checker in your head is ridiculously unreliable.
Quote
rntz
@arntzenius
I don't like dependently typed languages.
I also don't like conventional statically typed languages.
I also don't like dynamically typed languages.
In all of them, the reason is the same: I spend too much time satisfying a typechecker instead of coding.
I am in tears, begging for C++ using Rust critics to read one—just one—history post on who developed Rust, where, and why.
I’ve said it before, and I’ll say it again: if you don’t need equational reasoning about your code, don’t use Haskell. Almost anything else is better across almost every other dimension.
If you don’t need GC-free safe memory management, don’t use Rust. There are at least half a
Quote
Tom Sydney Kerckhove
@kerckhove_ts
Every time I see some company quit Haskell, I think "No wonder you want to quit Haskell if you insisted on doing something THIS stupid with it."
So maybe the missing puzzle piece for broader Haskell adoption is the ability to make many really stupid choices and still make bank.
I feel sorry for Jonathan Blow at this point.
Quote
ULTRA/MAGIC
@ULTRAMAGlC
"Dependent types" are just #define macros with arithmetic. C programmers have been doing type-level math in the preprocessor since before 1970.
If someone uses these terms as though they were some new amazing inventions, they probably do not have much programming experience yet. x.com/Jonathan_Blow/…
Unpopular opinion: I liked the first "Reacher" movie with Tom Cruise, and find fans' inability to see past "but he doesn't look anything like the books' description" an utterly bizarre failure of the imagination. Especially since we got Tom Cruise, Rosamund Pike, Robert Duvall,
Quote
E n z o k
@enzok
Replying to @neontaster
They did great with Reacher, but they shat the bed with RoP, and made a mockery of The Wheel of Time. So, this could really go either way, but most likely it will end poorly.
Smalltalk is the OOP paradigm exemplar. I believe all programmers should try all paradigm exemplars at least once:
* Pascal for imperative programming (freepascal.org)
* Smalltalk for OOP (squeak.org)
* Kernel for untyped functional programming
Quote
sai
@texoport__
hi, i think it's a tragedy that i haven't given Smalltalk a real shot, don't you?
Replying to
I saw quite a bit of this in Old San Juan, Puerto Rico, and also enjoy it. It apparently is a kind of low-tech approach to climate control in tropical areas, with the combination of sun and breezes keeping the courtyard relatively consistently temperate.
Another disastrously bad take. What is it with designers of systems programming languages feeling the need to publicly advertise their ignorance of type theory and existing work?
github.com/typelevel/squa
github.com/typelevel/squa
Quote
gingerBill
@TheGingerBill
Top tip: Don't try to model "units" (e.g. kg, s, lb, etc) in the "type" system of the language. Units are not types, and types are not units.
A trivial example of why it is bad:
f32 * f32 = f32
kg * kg = kg^2
1. "You can always remove one more line of code."
2. "There is always one more bug."
So, by induction:
3. "Any program can be reduced to a single line that doesn't work."
Quote
Kevlin Henney
@KevlinHenney
A number of years ago a team I was visiting was laughing that one of the other teams had developed software that only had two bugs: the first bug was that it typically failed to start; the second was that when it did start, it immediately crashed.
Your periodic reminder even "memory works the way C presents it" is false.
Quote
nhatanh
@alittletyper
Replying to @MathRestaurant
Not really, it's just... POSIX standardized syscall interface as C API, and the next popular thing (win32 API) also provides itself as C/C++ API, so it's hard to escape C unless everyone decided to completely branch off from current OSes.
terms indexed by terms = functions
terms indexed by types = overloaded functions
types indexed by types = “generics,” “templates,” “parametric polymorphism”
types indexed by terms = dependent types
youtu.be/SWTWkYbcWU0?si
h/t
Quote
laron ladler
@Aron_Adler
I tried to explain dependent types to someone irl the other day and it was so much harder than I was hoping it'd be
Feels like the kind of thing that would be much easier sitting at a computer together with plenty of time to build up the concepts and intuitions from scratch
Replying to
No, but fandom in general is. Like, a couple can be fans of a particular actor or director, and a couple can be fans of a particular game studio, genre, franchise...
My point is, the idea that video games are this weird, solitary, masculine activity is out of date.