Hey now, I'm implementing interpreters badly and realize it.
Sadly it is "too complex" and it was suggested to just go with a simple yaml file. :sob:
@typeholes.bsky.social
Interested in all things programming, cooking, cats. Love to explore new programming languages and build dev tools.
Hey now, I'm implementing interpreters badly and realize it.
Sadly it is "too complex" and it was suggested to just go with a simple yaml file. :sob:
Teams solves this problem in the worst way possible
10.11.2025 22:19 โ ๐ 1 ๐ 0 ๐ฌ 0 ๐ 0When we got an obnoxious assignment in my high school latin class, I shouted a long rant of latin profanity and insults at the teacher. I got an A for the course and excused from the assignment:)
07.11.2025 02:48 โ ๐ 4 ๐ 1 ๐ฌ 0 ๐ 0Would it be cheaper to let it double up and then do a post pass to collapse consecutive ones?
29.10.2025 23:57 โ ๐ 0 ๐ 0 ๐ฌ 1 ๐ 0We need to open a channel for back porting from Ark* into TS
29.10.2025 00:07 โ ๐ 1 ๐ 0 ๐ฌ 0 ๐ 0Using non continuously differentiable functions like floor is no fun at all.
21.10.2025 02:25 โ ๐ 0 ๐ 0 ๐ฌ 0 ๐ 0You could also ask if there are any affine units that have multiple useful torsors. The obvious torsor for points is vectors, but we could also choose pairs of rotations and scales relative to some fixed point. I think there are use cases for the latter
18.10.2025 02:39 โ ๐ 0 ๐ 0 ๐ฌ 0 ๐ 0The deltas are sometimes called torsors which seems to get more search hits than affine units.
Loose thinking, but if affine units are just choices of identity elements then they are a bunch of distinct single things so it makes sense that you can't combine them
TFW when you hit the documentation and where the documentation should be is two links to video tutorials :(
08.10.2025 23:05 โ ๐ 1 ๐ 0 ๐ฌ 0 ๐ 0I'd really like to see your results and methodology.
07.10.2025 23:53 โ ๐ 0 ๐ 0 ๐ฌ 0 ๐ 0Maybe "technique x allows new developers to contribute without the overhead of building a complex mental model"?
02.10.2025 03:43 โ ๐ 0 ๐ 0 ๐ฌ 0 ๐ 0I've said that a few times and honestly meant it, but generally I agree. I tried shifting to "your ability to keep track of ... Is really impressive" but that ended badly in cases where things were not kept track of.
02.10.2025 03:43 โ ๐ 1 ๐ 0 ๐ฌ 1 ๐ 0Seems like the perfect occasion to enjoy Surstrรถmming and durian salad
01.10.2025 20:41 โ ๐ 0 ๐ 0 ๐ฌ 0 ๐ 0Same. I reach into my pocket to grab it but it's not there
23.09.2025 19:14 โ ๐ 2 ๐ 0 ๐ฌ 0 ๐ 0This is the skeet making people seethe, isn't it?
13.09.2025 12:45 โ ๐ 0 ๐ 0 ๐ฌ 1 ๐ 0If you're able to ask this with a straight face then I'm seriously envious.
12.09.2025 15:42 โ ๐ 1 ๐ 0 ๐ฌ 1 ๐ 0Adding alt text on mobile ๐
11.09.2025 21:22 โ ๐ 1 ๐ 0 ๐ฌ 0 ๐ 0I've done thousands of interviews. Seen great candidates rejected because they haven't memorized some API or fumble code under scrutiny. The successful hires I've made had one thing in common: we can have a conversation about an interesting project
11.09.2025 03:01 โ ๐ 1 ๐ 0 ๐ฌ 0 ๐ 0Companies should change how they interview unless they can preprocess all the important information into the necessary technical details for the organic machine they're trying to hire.
11.09.2025 03:01 โ ๐ 0 ๐ 0 ๐ฌ 1 ๐ 0This is gold. I've got a bad case of CRS (can't remember stuff) but make up for it by rapidly deriving from principals. The best authors follow your advice which helps me and anyone not intimately family with the content.
11.09.2025 02:44 โ ๐ 1 ๐ 0 ๐ฌ 0 ๐ 0A dynamic vue component that takes a match clause and exposes props for the component to use for each case.
10.09.2025 17:46 โ ๐ 0 ๐ 0 ๐ฌ 0 ๐ 0I resisted this for 20 whole hours but the twitch will not stop
x^(a+b) = x^a * x^b
x^ab=x^a*x^b
---------------------
x^(a+b) = x^ab
sorry (in both the lean and apologetic meanings)
As a professional in SE who is unskilled but passionate about math this makes me scream every day. Literally.
I get my revenge by subjecting my coworkers to an excess of notation abuse in crazy EDSLs.
Honestly, nerd sniping David is too easy to qualify as a sport. Someday soon I'm going to wake up and my example's type will infer as 1|2.
06.09.2025 17:11 โ ๐ 3 ๐ 0 ๐ฌ 0 ๐ 0Think about they types you can map over the input and output of a->b
The input needs ?->a no matter what you need to get to an a
the output needs b->? you can use the value in b or ignore it, and you are free to return anything at all (including nothing)
Most things involving a flipped arrow feel unintuitive to me but make sense when I think it through.
My line if thought is the left side is taking a value. The right side is giving a value. The are opposite (dual). Any mapping has to preserve that opposite nature.
ugh, I meant >0 in Zplus but you get the idea.
I'll be seriously impressed if you can make this a type error with reasonable check times.
Z* bites me again. Let's try in Z+
import { type } from "arktype"
const Zplus = type({n: "number%1>1"})
const N3 = type({n: "number%3"})
const Ngt = type({n: "number<3"})
const Thing = Zplus.and(N3).and(Ngt)
const out = Thing({
n: 0,
})
Done
import { type } from "arktype"
const N3 = type({n: "number%3"})
const Ngt = type({n: "number<3"})
const Thing = N3.and(Ngt)
New KPI landed?
27.08.2025 17:09 โ ๐ 1 ๐ 0 ๐ฌ 1 ๐ 0