Maya :3 retweeted
does anyone know a typesetting system that supports patches? one that remembers where it put all the line/page breaks and can redo specific modified pages in a way that is seamless with those around it?
Replying to @serialismus
nie jest legalne, jedynie zdekryminalizowane. tbh nie rozumiem czemu pierwsze skojarzenie każdego to zioło... government.nl/topics/drugs/t…

ALT Toleration policy regarding soft drugs In the Netherlands, it is against the law to possess, sell or produce drugs. However, the Netherlands has a policy of toleration regarding soft drugs. This means that the sale of small quantities of soft drugs in coffee shops is a criminal offence but the Public Prosecution Service does not prosecute coffee shops for this offence. Neither does the Public Prosecution Service prosecute members of the public for possession of small quantities of soft drugs. These quantities are defined as follows: no more than 5 grams of cannabis (marijuana or hash); no more than 5 cannabis plants.
Replying to @thingskatedid
the general recommendation I've heard is "take a look at the DIVA diagnostic manual and see how much it's calling you out"
(if you find yourself unable to focus on the questions that's an automatic pass /hj)
Replying to @lethalbit
b-but... what if there's null bytes on that floppy? 🥺
Replying to @rose_bohrer @__phantomderp
Wouldn't that just mean sizeof(word_t) == 1?
Replying to @arntzenius
if you call your HOF with f as argument, make f an input to the incremental computation, have an intermediate query that asks for the value of f at a specific point, invalidate the input f every time you call your HOF.
this does get ugly if you have reentrancy though
Replying to @arntzenius
there is a certain isomorphism between the implementation strategies, though
Replying to @meithecatte @arntzenius
question is, whether you're memoizing because the HOF does expensive computations, or because you're applying it to expensive functions. but for the latter you'd just put the memoization on the first-order function, no?
Replying to @meithecatte @arntzenius
okay, I might have brained myself in how close the thing is, because I'm not sure they actually use this with higher-order functions anywhere, but it's basically the same idea – record the queries your HOF makes of the passed-in function, run them in the same order and compare
Replying to @arntzenius
this sounds like salsa: lib.rs/crates/salsa
used in production in the Rust compiler
I think I learned about it from this talk (there's a chapter titled "How Salsa works" 27 minutes into the video): youtube.com/watch?v=N6b44kMS…
Replying to @lethalbit
perhaps this could tell you something about where it's stuck github.com/tokio-rs/console
Maya :3 retweeted
Each major Game Boy hardware component's buggiest feature:
PPU: The window
APU: Channel 1 frequency sweep
CPU: The STOP instruction
OAM: Remaining idle without corrupting itself
Replying to @Myriachan
on the Game Boy, the term of choice is "echo RAM"
Replying to @CihanPostsThms
Okay, but if you don't have a proof, there's absolutely no reason to call the value your program outputs "BB(47)".
Replying to @meithecatte @CihanPostsThms
There is no algorithm to simply compute the value of BB(47), by construction.
Replying to @CihanPostsThms
There is no deterministic process of finding the value of BB(47). You just analyze all the machines and try to prove they don't terminate or find a way to execute them faster than the naive algorithm. You can apply various techniques to discharge the easy cases, but still.
Replying to @CihanPostsThms
well, if the Goldbach Conjecture is false, then BB(47) is lower-bounded by (a function of) what the smallest counterexample is.
Replying to @CihanPostsThms
but wouldn't we have to decide the goldbach conjecture to even find the value of BB(47)?




