« first day (3966 days earlier)      last day (50 days later) » 
00:00 - 13:0013:00 - 00:00

12:01 AM
Q: AoCG2021 Day 8: Delivery Corrections

BubblerPart of Advent of Code Golf 2021 event. See the linked meta post for details. The story continues from AoC2015 Day 3, Part 2. This challenge was kindly contributed by Wheat Wizard (Grain Ghost). Santa is delivering presents to an infinite two-dimensional grid of houses. The delivery begins deliv...

12:22 AM
@DLosc ¯\_(ツ)_/¯ (assuming Unicode is allowed)
12:34 AM
@rues how do you like windows 11 so far?
It's going good
asking because I'm considering updating
would you recommend?
I had a terminal in a remote window in VS Code crash on me once, but idk if it's a Windows 11 thing
@lyxal I don't want to recommend it to you because I've only been using it for a few days at this point and I have no idea if Windows 11 explodes in rare cases :P
fair enough
I'll wait till you know more
I haven't encountered anything bad yet, though, so I guess you could go ahead and update if you really want. You have 10 days to downgrade to Windows 10
Do you use GrabCAD? Some people I know were having problems with that on Windows 11
12:37 AM
no, I don't have anything CAD lol
(fyi I lost 70 rep from that user removal; I only just got around to checking)
This is what the home screen or whatever it's called looks like for me (I just chose a solid color but there are nicer themes if you want)
@Neil fixed
Oh, and even apps such as snipping tool and control panel look nicer now
ooh that does look nice
12:42 AM
It looks very Mac-y, all the windows, buttons, etc. are rounded
1:00 AM
Sandbox posts last active a week ago: Silly Sentence Generator, Implement a BrainFlump interpreter
@OldSandboxPosts sensible paragraph destroyer, deconstruct a skullflump(?) compiler
A paragraph is the opposite of a sentence?
it doesn't have to be the exact antonym
just relatively opposite
@OldSandboxPosts Sensible Pardon List, Specify the HeartBlint babbler
1:35 AM
but the opposite of generator is consumer
So it should be sensible pardon consumer
1:51 AM
Q: Choose the best colors to re-color an image

Nigel BessGiven some set of pixels large set of pixels (R,G,B, no alpha), a cost function, and a number n, I would like this algorithm to return n different colors such that the number of pixels that map to each color has minimal variance. I will explain what that means: Mapping: A pixel 'maps' to the colo...

2:11 AM
@lyxal Just so you know, I get an email every week with these :P
@cairdcoinheringaahing I know :)
@lyxal caird, it’s me, rues, from one week in the past. If you’re getting this, it means Earth’s screwed. If you’ve invented time travel yet, please come back into the past to get me
Aaargh wrong msg
I’m not gonna bother editing that
@NewPosts well that challenge changed a bit
Wow how did that happen
@rues If I ever invent time travel, my first act will be to create so many paradoxes :P
Physics is only fun if you break everything involved :P
2:23 AM
@NewPosts I don't know how to explain what's wrong here
Since nobody else has posted the LYAL message, I guess I will
Welcome to the eighth Learn You A Lang For Great Good! Today, we'll be learning Lost, a non-deterministic 2D language made by Grain Ghost. Feel free to ask questions about Lost, post CMCs ("LMC"s) related to the language, and just discuss the language in general. Try it online!
CMC: write the program that does nothing, then ends
or ends and then does nothing, the order doesn't matter
oh boy this is insane
tbf in lost it's not entirely straightforward since you need to hit the safety before you can actually terminate
...that's why it's a CMC
2:39 AM
wonder if it's actually possible in less than 5 bytes
feels like it should be but then i actually think about it and the answer looks more like no lmao
yeah it needs %, @, and another dimension to catch IPs
i'm more thinking of some kind of arrangement of
with one `%`, one `@`, and one control flow op, but the control flow op can't be both of the bits that sticks out so no matter what you have a trivial infinite loop
hm why won't a single @ just work
the safety starts on
2:45 AM
@ only terminates the program if the safety is off (i.e. the ip has hit a % more recently than a #), because making deterministic programs would be a hell of a lot harder if the ip could randomly start on top of @ and end execution instantly
so I have \/,%@
oh lol, i took it seriously when someone said earlier that the language was "Loss"
i have what bubbler has; didn't occur to me that you actually can do it in one dimension lol
sorry I can't use TIO, but I'm sure Bubbler's solution is correct
@UnrelatedString if you mean my solution, the comma means newline sry
i'm just running everything with the -QA flags
oh that makes sense
2:46 AM
No, you can't do in 1D because the IP might start vertically on a non-mirror
yeah because you literally can't do it in 1d
@% halts with probability 1/2
are there any lost solutions that use #? i don't remember ever using it
rats, I can't even exit by erroring because I can't find a way to error
good question
i can sort of see how it could come up in a very carefully golfed layout
but that doesn't mean it actually has happened
2:52 AM
CMC: produce an error (out of memory errors give only half credit)
so i assume the "scope stack" is the only thing that's printed?
i think it's just the main stack that's printed
scope stack is handy extra storage
yup. be careful of pushing unwanted zeroes though
@JoKing that’s not deterministic tho, since the IP can start on the safety and go right to the halt
2:58 AM
i mean, i wasn't aiming for deterministic, just that it produces an error. i'm already relying on the -A flag, so why not on the -Q. But fine, i can move the % to the start: Try it online!
oh yeah true you can just consider it as specifically a flagged submission lol
@JoKing oic
what are the flags in this language
ah -h
> Klein Interpreter
3:06 AM
CMC: add two numbers
Probably doable because extra addition is noop, but multiplication will be frick
i;m trying to do this random number challenge
why's this pushing an extra zero sometimes
What if it spawns on the safety going towards the halt?
wraparound i guess
getting this to display exactly one number is.. something
when all else fails you can make a padded to hell latin square
it won't ever even be close to shortest but it does work
3:25 AM
how do you even take input??
the flags say it is after the Source, but i guess you can't do that on TIO
cli args
looks like it's automatically added to stack
oh, you have to do it before the other args
luckily wheat wizard has a few answers in this lang
@emanresuA 9 bytes. I feel like there should be a byte that can be removed, but i can't find it
CMC: get the sign of a number (-1,0,1)
3:37 AM
input is really hard
'cos you have to be careful not to delete it?
or do something like - and invert the output
i mean
lost allows inputting any int
3:52 AM
is rn lyal?
@JoKing ...why did i get stars for this
@PyGamer0 get lost
oh ok
i can't find any previous lost answers that take input apart from this, which is basically the same as my previous answer
4:10 AM
CMC: Add one to the input (restriction: only positive)
... is deterministic Lost even Turing Complete? If you're taking input you can't use ?, since that might discard the first number, therefore there's no branching
@GrainGhost ^
@cairdcoinheringaahing Imagine not configuring your bot to auto-ack pings :p
CMC: Multiply two numbers
I have an idea for this one
Uh no
i don't think it's possible
@JoKing Next LYAL: I II Ii I_
4:25 AM
it mostly works but I can't remove exactly four paths that halt immediately
I had this
@cairdcoinheringaahing .:|:;
@RedwolfPrograms The key reasoning behind not doing that is that I don't know how, and can't be bothered :P
As good a reason as any lol
Why learn something when you can not? :P
@cairdcoinheringaahing She says, while at university, specifically to learn things :P
signum looks possible in deterministic lost
4:33 AM
In case anyone was wondering why their /activity page looks different, the new responsive redesign has been officially rolled out to all sites
ha, i was scared of bringing that up in case it had happened forever ago and i had missed it
> 500
up votes
not a fan, the padding and size is stupidly large
CMQ: remove LCT (LCC in Jelly) from flax?
@JoKing Keeping an eye on MSE has its perks :P
4:36 AM
@JoKing i agree
@PyGamer0 Lowest Common Tractor?
@cairdcoinheringaahing no: Largest Commercial Train
that rammed into the dataserver
@JoKing Everything's larger, but I don't mind the fact that the new info is available
I dislike the fact that the "next badge" and "next tag badge" bits are crammed together
hey guys what are articles?
it 404s but it's on the profile pages
It's for collectives on SO
4:40 AM
Yeah, it's big and ugly
I think they basically standardised profiles across the network, which includes articles on non-SO accounts
CMC: Golf down an answer you posted at least six months ago
How is there no info on the reputation section on the summary page? like no dates, no numbers, literally just green rectangles
impossible CMC, all my answers are perfect
@emanresuA people who joined <6 months ago: o_O
4:43 AM
Lost CMC: Print 1 half of the time and nothing otherwise
@emanresuA yeah i know
@JoKing mmmm, love me some green rectangles
Q: Articles on MSE?

LuuklagI noted the "Articles" tab here on my MSE profile: Does this mean that articles are coming to MSE as well (which wouldn't really serve a purpose IMHO), or is this a bug and should this only be visible on SO?

@JoKing FWIW, you can hover over the green rectangles to get more info
ah ok, i was trying to click on them, and the tooltip doesn't show up after you do so
i guess someone's already reported it: meta.stackexchange.com/a/372603/381197
Is SE updated?
4:46 AM
my rep changed yet again
1135 => 1121, minus 14
i think the quantum boolean thingy just got yeeted
That's an interesting change
Three downvotes and an upvote I guess
it's +2/-3
but my rep was 1135
hold up
4:48 AM
@Niko Yep, Community deleted it an hour ago
my math just went back to elementry school for a moment
It's understandable, math with numbers is the worst kind of math
I have a idea of fixing that quantum question
I think doing that transpiler thingy only will make most language possible to solve
4:59 AM
@Bubbler 9 bytes
that was more randomly rearranging the code than it was actual thought
5:17 AM
if i remember, did lyxal setup a part of his web to allowed vpn-alike website view?
like using one site that embedded another site?
CMP: Youtube to .mp3 converters?
This change makes we want vomit. It's absolutely horrible. I would rather remove my teeth with a screwdriver then look at it. Please revert and throw this garbage change into the trash bin. I wish I had something nice to say, about the change, but the simple fact is I actually hate it. — Ramhound 4 hours ago
See, I prefer the old activity page, mainly cause I get more info at a glance. But this is such an overreaction that I understand why SE is hesitant to post feedback posts on MSE
Should interleaving a string and an array return a string or an array?
5:35 AM
6:24 AM
@Niko that was indeed me
It was just embedding OEIS sequences
that's the equivalent of just searching the OEIS for sequence 1,2,3,4,5
6:40 AM
@PyGamer0 just pip install youtube-dl
^ any way to make that loop O(n) or it cannot be O(n)?
technically its not O(n²) as it recurs
@AaroneousMiller looks like it's been decided anyway, but I was gonna suggest we do Lost thiis time anyway
i think you actually can construct a quadratic worst case
or actually i guess we need to define what the input size actually is
but yeah it won't do too hot on something that's just one longish list nested super deep because it'll keep traversing the whole thing repeatedly
you could also pass the list you're building around
6:58 AM
^ is that an improvement
i strictly want a list
can just put a wrapper around the generator function then
should be
I thought of this but wrapping the generator should also work
Both are O(n) where n is the length of the resulting list
@Bubbler the non-generator solution i was thinking of was pretty much that except with an optional argument instead of an inner helper
def flatten(L,buf=[]) doesn't work btw
7:03 AM
wrapper would be like this
can use buf=None then just substitute manually
i never knew about yield from
what does it do?
7:16 AM
ah my github token expired
just set it to never expire
@PyGamer0 it's sort of like just looping through the thing you yield from and yielding each of those
@Razetime i set to expire on 1st december 2022
however, that's only for very elementary applications. yield from does way more than for x in a: yield x could ever hope to do
cause if i switch computers i would need to make a new one
7:20 AM
see this SO post for more info
@hyper-neutrino ok thx
8:07 AM
@cairdcoinheringaahing Wow, the announcement is now at +51/-51 O.o
> After receiving and incorporating feedback [from] the community

Press X to doubt
i really dislike chat markdown
@cairdcoinheringaahing net: ¯2
*All info was correct at time of posting :P
@JoKing One thing to note is that the feedback before earlier today was based on the video, not the actual deployed product
they look very different from each other
Most upvoted post on the linked mobile view deprecation question: "Can we increase information density on the responsive site on mobile?". I think they took this to mean "Can we decrease information density on the responsive site on desktop?"
I really cannot understand why you would do anything other than start with the current design and only modify it to be responsive. — pxeger 5 mins ago
Why was it necessary to start from scratch?
This has resulted in a large number of obvious bugs (which should've been caught before going live) but wouldn't have happened at all if you started with the current one
8:16 AM
It's been fun to watch this go from score 16, 3 answers to -7 with 74 answers, all within 12 hours :P
The old design was already perfect on desktop, so no changes should have been made to it!
yeah, making the old design work on phones would've been a better approach
i dunno about everyone else, but it still looks pretty bad on mobile. seems like the page width is still too wide
There's too much empty space on mobile, but its nicer than the old "just give them a zoomed out desktop view"
uhhh, i mean, it still is the zoomed out desktop view
8:25 AM
Its more of a zoomed-in desktop view
The mobile version (for me at least) needs a bit of tweaking to remove all the empty space, but is otherwise better than it used to be
don't judge my navigation bar
oh, looks better on chrome, maybe it's just the browser
@JoKing Better than on desktop, cos you can actually see stuff.
i mean, i can't read the words, but sure
@cairdcoinheringaahing ah got it working. dunno why it was broken. i mostly agree with this
whats the use of attrdict in jelly souce, it is just for storing the arity and call of atoms?
8:41 AM
It's a type that allows object.method to be syntax sugar for object["method"]
thats it?
and what does it store, just the info about atoms (their arity/method to call)?
It stores whatever you want, but it's typically used as attrdict(arity = n, call = function, ldepth = l, rdepth = r)
whats ldepth and rdepth?
The depth that the function vectorises to on the left and right
left and right, the arguments?
8:45 AM
If you have e.g. f that has ldepth = 1, rdepth = 0 with [[1,2],[2,3]] f [[3,4],[5,6]] you get [[1,2] f [[3,4],[5,6]], [2,3] f [[3,4],[5,6]]]
hm what if ldepth = 2, rdepth = 1?
[[1 f [3,4], 2 f [3,4]], [2 f [5,6], 3 f [5,6]]] I believe
so are ldepth and rdepth constant for a certain atom?
Yeah, if they're not given, it vectorises to full depth, as in this question
@cairdcoinheringaahing Nope, if not given they default to not vectorising, and a depth of zero indicates full depth vectorisation
9:20 AM
@JoKing If by deterministic Lost you mean a subset of list programs that are deterministic you are mostly correct, there is no way to take input outside of a few specific functions. But TC does not require input, it requires a mapping between programs that preserves halting behavior. It is my belief that Lost is in fact TC.
Multiplying inputs is provably not possible.
I have been considering a modified version of lost in which types of input are possible.
You can also give very weird input formats.
@GrainGhost Might be possible if you can encode both into one number and take that multiple times
decoding would be the tricky bit though
Or just some padding on the input would even do it I think.
Is there any meta consensus that allows taking an input multiple times?
No, that would be weird imo. But for lost specific CMCs / challenges I would say that using odd formats is fine. Esp. for learning how to do things.
i think convoluted input formats tend to only be prohibited to the extent that they encode extra computation
so it's probably fine for a lost program to require some large-but-constant amount of padding
but actual redundancy would sort of have to reflect something deeper so it's probably out
9:28 AM
CMC: Multiply two numbers in lost with the least amount of padding.
9:38 AM
is there a term for a language that can map all possible inputs to outputs?
What do you mean by that?
like, you can write a program for any mapping of a set of inputs to a set of outputs
If I'm understanding it's not possible, there are more output input combinations than there are finite strings.
uhhh, what if the input/output is finite?
Ah ok, doable. Idk if there's a name for that.
9:45 AM
yay I managed to golf my efficient sum powers solver down to less than my brute force solver
this is usually what i really mean when i think TC (i know i'm wrong)
10:03 AM
In terms of the normal hierarchy it's normal, but it has this extra restriction, which is basically just a encoding restriction. I think for theoretical computer science that stuff is just usually ignored.
Made a userscript which makes BQN submissions easier.
(Don't ask whether those are my arms, legs, tentacles or all three)
10:26 AM
i'm lost
@LeakyNun no, you're Leaky Nun
i am bored
6 hours ago, by PyGamer0
CMQ: remove LCT (LCC in Jelly) from flax?
10:53 AM
@GrainGhost 2 padded 1s. I can't quite figure out a single padding (yet)
11:26 AM
@Razetime Nice. Installed. You may want to add none to the fence so SE doesn't try to guess the language.
11:47 AM
holy crap I just found a reference to ais523 on a TAS run of a brain age game
at around 17:17
@GrainGhost 42 bytes, one padding (though it doesn't work when multiplying a zero)
i had to download the interpreter to debug this, did you know there is a single space at the beginning of line 23 of interpreter.py?? fun
The interpreter is ... bad.
12:20 PM
@lyxal wow what the hell
i've seen thsi twice and i didn't notice
CMC: print "Lost" in Lost without trailing newlines
That's not going to happen if you use the python interpreter.
Is possible?
No the python interpreter always prints a trailing newline.
uhhh i never learned Lost
What is paddings in Lost?
12:32 PM
We just mean that you put some junk on top of the input before executing.
It's to prevent accidental damage to the input.
ok so
When the IP moves randomly?
Well it doesn't so much move randomly as it starts in a random place.
The first instruction it executes could be something that just eats the input like ?.
@GrainGhost oh sorry
@Adám will do
update coming up soon-ish
what does -A Flag do?
12:35 PM
It toggles whether the output is printed as a string or as integers.
The default is integers.
space separated
12:37 PM
@GrainGhost Is there program with infinite output?
No output is done at the end.
So it's not possible.
If there were a "output now" command it wouldn't actually be very useful since you can't undo output. It would almost always have a chance to output stuff you don't want near the start of the program.
so... the IP Wraps?
Yes, like a torus.
Although ... you could link output to safety ... then you would be able to control it a bit ...
ok so i'm making a program that outputs 9
12:41 PM
Yeah that's a nice starting point.
without -A flag
Yeah, definitely easier.
oh no
sometimes outputs 9 9
i'm fixing that
if you add the -V flag it will run all the outputs.
*give up*
12:46 PM
( will let you get rid of the extra 9.
@GrainGhost yeah but i still can't
hm. Well first putting % next to @ is risky since it can start on % and then immediately exit.
So your initial program where you had stuff in between them had the right idea.
ok so i give up
12:48 PM
Well pygamer posted a working solution.
It's very close to what you started with.
the -V flag is very useful
but i still can't
Try and imagine % as the start of the program. It can do a lot of nonsense first, but once it gets to % then your just need to clean up and actually start the program.
LostCMC: Print the factorial of input
@PyGamer0 oh no
12:52 PM
Going to need some padding for that.
@GrainGhost can i do first in ><>?
then transpile to Lost?
LostCMC: Compute input - 1
Probably a good idea. Klein 000 is more similar to lost than ><> is, but having a sketch before diving in is a good idea.
The only things I currently know can be done without padding are: cat, output a constant and add two numbers.
how to put input number to stack?
Which stack?
12:55 PM
Input starts on the stack.
Which in retrospect may have been a poor design choice.
Why not working? Try it online!
random thing:Try it online!
12:59 PM
input is via command line, e.g Try it online!
00:00 - 13:0013:00 - 00:00

« first day (3966 days earlier)      last day (50 days later) »