The Garbage Collector

Rydwolf Programs

22:00

Ohhh that makes sense

user

@naffetS That's a float thing, right?

@RydwolfPrograms ye

Rydwolf Programs

^^

Any lang following IEEE-754 will have those Infinity and NaN behaviors for its floats

Bbrk24

I did notice that different languages handle sqrt(-0) differently though

and not all of them even handle it the same as (-0)**0.5

Dannyu NDos

And that's why I must replace IEEE-754.

Rydwolf Programs

Might be controversial but I really love how floats handle infinity

Dannyu NDos

22:04

Arbitrary-precision real numbers are what I'm trying to achieve.

Rydwolf Programs

Problem is that those are big and slow

For some applications you can use sympy and/or an infinite iterator approaching the right answer, but for lots of applications you're doing millions of computations in a row and you can't have a number type that takes ten times longer and gets bigger every time you use it for those

Dannyu NDos

Dang it, Cantor. It's thy fault that real numbers are uncountably many!

Bbrk24

@DannyuNDos What would Math.PI be then?

Rydwolf Programs

Depending on the way you approach it, either a symbol representing Pi, or an infinite series approaching pi

Dannyu NDos

@Bbrk24 Some exact representation of π.

Rydwolf Programs

22:07

My upcoming golfing language will use the latter

I'm working on a library called RIPR

(Rust Infinite Precision [something], what it stands for doesn't really matter I just think RIPR sounds totally awesome)

Bbrk24

@DannyuNDos So then you can beat the world record for most digits of pi with print(Math.PI)

(I'm being facetious but do you see what I'm getting at?)

Dannyu NDos

@Bbrk24 Nah, that won't halt.

Rydwolf Programs

Typically the way you do it is you store a symbolic representation then you generate a number of digits from that once the computations are done

So print(pi) would either print with a certain default precision, or it would maybe print indefinitely, but there's no reason it wouldn't work

@RydwolfPrograms Although the infinite iterator approach doesn't sound like this, in reality the in-memory representation is basically the same

Dannyu NDos

I'm just hoping that I can optimize the real-number computation, owing to the computation class they belong.

Dang it, Chaitin's constant. You're RE-complete!

Rydwolf Programs

RIPR's gonna have complex numbers too ofc

Reals are so 1544

user

22:15

@NewPosts I wish I could nuke questions

Seggan

@Bbrk24 ~~Use Rust~~

@user tbh that wouldnt be fair

user

I know, I know

Bbrk24

Does Rust support struct int24_t { int32_t value : 24; };?

because that's like, the only reason I chose C++

Seggan

as in bounded to 24 bits

Bbrk24

yes

Like Swift will let me implement my own type but I'd have to implement all the operators myself

Dannyu NDos

22:18

Dunno, struct int24 { lo : int16, hi : int8 }?

Bbrk24

yeah but then you have to handle wrapping manually

Seggan

^^

Rydwolf Programs

@Bbrk24 Does C++...not require that? How does that work?

Seggan

@Bbrk24 why are you restricting to 24 bits again

Dannyu NDos

Integer overflow is undefined behavior in C/C++.

Bbrk24

22:19

@RydwolfPrograms When you write int32_t value : 24;, you're telling the compiler to truncate it to 24 bits on every assignment. This is meant to support bitfields when you have multiple values packed into a byte/word, but

Rydwolf Programs

Ohh didn't see the : 24

Yeah I don't think Rust has that unfortunately

Dannyu NDos

Do bit fields work for pointers tho?

Bbrk24

@DannyuNDos github.com/bbrk24/Trilangle/blob/master/int24.cpp#L5

user

@DannyuNDos Just use void*

Wait you shouldn't even need void

You mean pointers to weird-sized structs? Those should work normally

Bbrk24

void* why: 3;

that's probably illegal

Dannyu NDos

22:22

Wait, I forgot why I need this. :(

user

Yeah I don't see why you should be able to restrict pointer size

Bbrk24

I also have a bitfield of booleans because if I don't do that it takes up seven bytes

Ginger

agh 899

Bbrk24

I've hit repcap four days in a row

Ginger

I'm 4 rep away from the cap

Bbrk24

22:26

I also find it funny when the repcap cuts off mid-upvote and so you see a notification that says [+4]

Rydwolf Programs

I mean I got a +4 without repcapping

Ginger

ah yes, an upv

Rydwolf Programs

3 downvotes and 1 upvote

New Posts

0

Q: What are the pros and cons of supporting Tail Recursion Optimization, but not Tail Call Optimization?

Tail Call Optimization allows a function call as the returned value of a function to be optimized to a goto, preventing the stack from growing. Among other things, this allows infinite recursion without worrying about the call stack size. While TCO applies to any function call performed as the fi...

Rydwolf Programs

Tryna ask fewer syntax questions and get more technical

Seggan

22:39

@Bbrk24 i only did 2 :(

@RydwolfPrograms yeah thats why im asking Rol stuff

Bbrk24

@mousetail I've written a couple hundred lines of code just creating the types necessary to support your algorithm

I haven't even started yet

why can't I initialize a queue of pointers as { nullptr } :thonk:

okay so I can a deque but not a regular queue

Rydwolf Programs

Looking like I might not repcap today rip

Bbrk24

std::queue<std::vector<instruction_pointer>*> fragments(std::deque<std::vector<instruction_pointer>*>{ nullptr });

dammit C++

wait I can make this a litttle shorter

auto fragments = std::queue(std::deque<std::vector<instruction_pointer>*>{ nullptr });

Rydwolf Programs

CMM: I've been thinking we need some sort of way to keep track of what questions we'd like better answers to

Since we have lots of list-style questions, the no-accepted-answers ones aren't necessarily a good list of them

Seggan

@RydwolfPrograms welp, i used up all my votes at least

Rydwolf Programs

22:52

Which means that lots of questions are going effectively unanswered, but invisibly so

E.g., this

Bbrk24

actually do I use std::list or std::deque

actually no I have to use std::vector because indexing

Rydwolf Programs

Make sure not to stream of thought pls

Oh on that note, should we have a chatiquette like TNB?

lyxal

23:15

@Adám as a matter of fact, no

No headaches there mostly because it's designed around the way I think lol

Rydwolf Programs

Lyxal's brain is canonically stack-based

user

lyxal is stacked

@RydwolfPrograms Let's also be less afraid to edit/close/downvote questions

Rydwolf Programs

*Unless they're mine

Michael Homer

@RydwolfPrograms I think this is fundamentally a solid question, but the pros & cons framing does it no favours. Something like "what are the trade-offs made by supporting only tail recursion optimisation" invites judgement & analysis that can weigh those factors instead of just enumerating them

Rydwolf Programs

"Trade-offs" and "pros and cons" are kinda synonyms

But yeah I can go do that ig

Michael Homer

23:24

That's why I suggested it!

Rydwolf Programs

Oh right lol

Michael Homer

But some framing (maybe not that one) that asks people to evaluate what they're saying I think leads to better answers

Even "is it worth only supporting..." could suffice, depending on what you want from it

Rydwolf Programs

"worth" implies a judgement based on weighing the two sides tho, which could add subjectivity

user

I kinda want to delete my extension methods vs UFCS question now

Rydwolf Programs

I like the trade-offs wording tho

@user I found it useful fwiw

user

23:26

Right, it's just not phrased the best way

Michael Homer

@RydwolfPrograms Good! An answer shouldn't just be a list of unweighted facts but make a case for how they weigh up and what you should care about

user

Kinda a low-effort question that I wrote in 5 minutes

Rydwolf Programs

True, but subjectivity isn't really encouraged here, so it'd be a bit of a fine line

user

@MichaelHomer Well, answers should also be try to objective. But I guess it does make sense to say "prefer <X> if <Y> is your use case, prefer <Z> if <A> is your use case"

Michael Homer

@user Yes, that would be a fine structure

Rydwolf Programs

23:28

Wha---I just got +9 from an upvote on a question?

user

Repcap? Downvote?

Rydwolf Programs

Yeah but wouldn't a downvote show it's from a different source?

OHHH repcap

user

Sometimes when you downvote others the notification acts weird

Michael Homer

You downvoted an answer earlier for a point

And now hit the cap

Rydwolf Programs

yep, that'd be it

Michael Homer

23:32

@RydwolfPrograms Also, frankly, answers that make a case should be encouraged and notwithstanding the "opinion-based" close reason they are all over the network: "use find -delete for this, because piping find to xargs is dangerous" is just as much presenting an argument

The problem is when the answer can only be "because I think so"

naffetS

23:49

i almost hit repcap again

missed by 12

Bbrk24

@mousetail This has made a lot more sense as I've been implementing it, but there's just one problem: two TJN instructions at the same location must be represented in the same line of assembly even if the IP direction differs, with one being a jump to the other.

Actually that shouldn't be too difficult to fix, I just change the map keys from (location, direction) to (location, instruction)

no that messes with the mirrors, where they must stay separate

It's like I have to add special case handling for the threading operator

naffetS

why is it in russian??

Rydwolf Programs

Well, we get access to 10k tools as soon as someone gets 5-10 upvotes

Then tag synonyms tomorrow for one or two people and the next day for everyone else

naffetS

I'm so used to 10k tools on cgcc that it seems weird not having them here

Rydwolf Programs

I tried viewing a deleted question the other day and the 404 or whatever was so weird

Transcript for