« first day (2922 days earlier)      last day (1904 days later) » 

ASCII-only
ASCII-only
12:48 AM
@Mego but still. could you not write a script to implement large neighborhood functionality? or even better, modify the existing neighborhood script?
@AdmBorkBork hmm @Dennis maybe you could add a AST printer?
 
Dennis
Dennis
Sorry, a what?
 
ASCII-only
ASCII-only
well. some way to somehow print out the train structure of a Jelly program
 
feersum
feersum
You can't build a structure with jelly.
 
ASCII-only
ASCII-only
:|
it should be kinda doable, since it actually parses :P theoretically all you would need to do would be to create another set of builtins, except they print the tree instead of do things
 
Dennis
Dennis
Jelly is interpreted. The structure of a link depends on how many arguments it is called with.
 
ASCII-only
ASCII-only
12:56 AM
does it depend on input args?
 
Dennis
Dennis
The main link does, yes. But you can still call it with a different number of args, from anywhere in the program.
 
ASCII-only
ASCII-only
oh :/
i guess it might be possible to print the structure of a link for every possible number of args. IDK how hard/useful that would be though
 
Anonymous
@ASCII-only I'd have to completely change how Golly loads rule files, since currently it assumes no more than 8 neighbors. That's more effort than I'm willing to put in for someone else's project.
 
ASCII-only
ASCII-only
@Mego since when? (wait brb)
@Mego the extension script has a 12-neighbor rule
sure you might have to change how it loads rule files
but the code is already there
oh wait what
 
New Main Posts
New Main Posts
1:20 AM
0
Q: Trapped Knight Sequence

Shieru AsakotoIntroduction Inspired by the very recent video The Trapped Knight - Numberphile, I came up with a challenge. The trapped knight sequence is a finite integer sequence of length 2016, starting from 1, and has the following construction rules: Write a number spiral in the following manner: 17 ...

code-golf number
 
 
1 hour later…
ASCII-only
ASCII-only
2:48 AM
@BMO if you're still here: 137?
 
lirtosiast
lirtosiast
3:15 AM
@ASCII-only well Jelly arities only go up to 2
 
Quintec
Quintec
When your professor typoes and asks "Prove P = NP" instead of "Prove P ⊆ NP" hmm
 
 
2 hours later…
New Bounties With No Deadlines
New Bounties With No Deadlines
5:40 AM
0
A: List of bounties with no deadline

Jo King200 rep each for answering unanswered questions I don't like seeing good questions go unanswered, so I'm offering 200 rep each for answering unanswered questions that have a score of 20 or above. As of now (January 29th), there are 3 such questions: Magic: the Gathering Combat Golf Halting Pro...

 
 
1 hour later…
lirtosiast
lirtosiast
6:55 AM
@ETHproductions Is Cubix Turing-complete?
 
 
4 hours later…
Adám
Adám
11:02 AM
Want to meet APL in NY? There are two upcoming events: one in NYC, and another one in Poughkeepsie. Sign-up is mandatory.
2
 
Bassdrop Cumberwubwubwub
Bassdrop Cumberwubwubwub
@NewMainPosts So I voted to close this question because someone already posted a similar question yesterday, the answers to that challenge can easily be ported to this challenge while only changing a few bytes. (See Kevins answer). However I just found out that Jonathan had never put his challenge in the sandbox. Which of the 2 is a dupe of the other?
4
Q: Trapped Knight Sequence

Shieru AsakotoIntroduction Inspired by the very recent video The Trapped Knight - Numberphile, I came up with a challenge. The trapped knight sequence is a finite integer sequence of length 2016, starting from 1, and has the following construction rules: Write a number spiral in the following manner: 17 ...

code-golf number
 
 
2 hours later…
AdmBorkBork
AdmBorkBork
1:22 PM
@Adám Sign-up is mandatory? But what if I don't wanna go?
 
Adám
Adám
1:33 PM
@AdmBorkBork It always bothers me on the London tube when they announce Change here for the Northern Line. But, but, I want to continue on this train!
 
New Main Posts
New Main Posts
2:04 PM
1
Q: Classic VCS ASCII Adventure

gwaughGrowing up, my first console game system was an Atari 2600 and I will always have a love for some of those games I so enjoyed as a child. Many of the graphics are still memorable, perhaps even iconic. It turns out that these sprites are very simplistic bitmaps, 8 pixels wide with variable heigh...

code-golf ascii-art
 
 
3 hours later…
Deadcode
Deadcode
4:46 PM
@Mego I felt that the comments on The Path Of The Wildebeest were very relevant to the problem, were terse enough to not take up much space, and had not devolved into chat. Why move them?
And I'm up to 210,918,965,248 steps so far – a wilderabeest with 1,2 + 1,3 + 1,4 moves still hasn't been trapped.
And by relevant to the problem, I mean that anybody reading that page would be likely to be interested in what they said. I will admit that the comments were not directly relevant to the solving of the challenge.
@EriktheOutgolfer Definitely agree on this; it would be very time-consuming to search for islands. And amazingly, it seems like it isn't even necessary. It seems like up to a very late stage, no islands are left... all squares up to a relatively very recent time are visited. (At least with the move variations I've tried so far.)
Well, unless the search were very limited. Such as just checking for a full barrier of a sufficient number of full spiral revolutions without even a single square of gap in them. In fact that might be even faster than what I'm currently doing, which is verifying that all squares up to the position-to-be-pruned are visited.
 
Erik the Outgolfer
Erik the Outgolfer
5:33 PM
lol everything's basically moot now the magic number was revealed
 
Deadcode
Deadcode
@EriktheOutgolfer How do you figure? That's just the 1,2 + 1,3 version.
Do PPCGers really have such short attention spans? :P
 
Erik the Outgolfer
Erik the Outgolfer
um, are you thinking about a general case for fairy pieces?
also... us PPCGers with "short attention spans" know what inspiration is very well...
 
Deadcode
Deadcode
I'm thinking about the general case for any set of moves, even those not implemented in any chess variant.
 
Erik the Outgolfer
Erik the Outgolfer
I guess brute force is necessary to find out... the amount of brute force, however, I'm not exactly certain :P
 
Deadcode
Deadcode
Right, unless somebody can figure out how to prove lack of trapping for any case.
 
Erik the Outgolfer
Erik the Outgolfer
5:40 PM
well... maybe if both of these conditions are met:
er, I thought of this again
maybe if the piece is stuck in a loop of neighborhoods (a neighborhood would be the set of positions the piece can move to at the next step and whether they've been visited before or not) then it will go on indefinitely
 
Deadcode
Deadcode
What do you mean by a loop?
 
Erik the Outgolfer
Erik the Outgolfer
whether a fixed point is met
for example, let's take the knight piece
it sits on position (x,y)
 
Deadcode
Deadcode
You mean a repetitive pattern that it will follow around each revolution, regardless of the size of the revolution?
 
Erik the Outgolfer
Erik the Outgolfer
yeah, a sequence of neighborhoods that is repeated at least twice
in order, with its "cells" in order too
that would of course mean that the piece doesn't follow a spiral numbering like in the challenge, that is, its moves are constant
(otherwise the squares aren't binary)
 
Deadcode
Deadcode
Okay now I don't understand. Are you talking about a different rule set?
 
Erik the Outgolfer
Erik the Outgolfer
5:50 PM
yes
because a rule set with square numbering is probably going to require brute force anyway
 
Deadcode
Deadcode
I thought you meant for example, the piece might weave around the spiral in a certain interleave each time, that manages to perfectly visit every square in a predictable way
I suppose it's plausible that could happen with some move set(s)
Although I think in most it will be chaotic.
 
Erik the Outgolfer
Erik the Outgolfer
so, transferring my assumption above to a numbered-square system, the neighborhood will also contain numbers
however, we don't care for the numerical values themselves, only for their ordering
 
Deadcode
Deadcode
@feersum Yeah, I thought a little about this and I think it'd result in slower code. It'd have to handle literal corner cases separately.
@EriktheOutgolfer But not sure why you bring that up?
 
Erik the Outgolfer
Erik the Outgolfer
so, in this case, we can say that two neighborhoods A and B are the same if the ordering if
1) A_x=V (value that denotes a cell has been visited), B_x=V (x representing a certain position in a neighborhood)
2) in sorted vectors of non-V cells of A and B called S_A and S_B respectively, S_A_i=A_x and S_B_i=B_x, i being an index
in ^, non-V cells have integer values, however, as shown, the values themselves don't affect the neighborhood, only their sort order does
 
Deadcode
Deadcode
I think it'd be more complicated than that... basically have to show that the pattern is invariant with respect to spiral size.
 
Erik the Outgolfer
Erik the Outgolfer
6:01 PM
I'm not assuming the numeric pattern is a spiral at all here
 
Deadcode
Deadcode
What kind of numbering are you thinking of then? Back-and-forth fill of a finite-width infinite-height rectangle?
or just CRT-style scan?
Width*row + column?
 
Erik the Outgolfer
Erik the Outgolfer
not a specific numbering, just an arbitrary numbering
wait a sec, it might be easier to understand with some ASCII art 
Neighborhood A |Neighborhood B
---------------+---------------
 .  1  .  2  . |  . 70  . 72  .
14  .  .  . 15 | 75  .  .  . 77
 .  .  N  .  . |  .  .  N  .  .
 V  .  .  . 12 |  V  .  .  . 74
 .  6  .  V  . |  . 73  .  V  .
N is the kNight piece
. is a location where the piece can't move to, hence it's disregarded
V is a visited square
now, let's assume that the indices of the cells of a neighborhood are:
 
Deadcode
Deadcode
Oh, so you're just talking about two immediate neighborhoods. I thought you were talking about comparing consecutive sets of revolutions. How will comparing two immediate neighborhoods at different time indexes help?
 
Erik the Outgolfer
Erik the Outgolfer
. 0 . 1 .
7 . . . 2
. . N . .
6 . . . 3
. 5 . 4 .
@Deadcode yes, those are two equal neighborhoods, the next step is to confirm that two consecutive (that is, with no other neighborhoods in-between) sequences of neighborhoods in the history are equal, according to the definition of equality I've provided above
 
Deadcode
Deadcode
Right, I was just going to say that
But the time difference has to be a function of the spiral size
 
Erik the Outgolfer
Erik the Outgolfer
6:11 PM
um... what?
 
Deadcode
Deadcode
And showing that the pattern would survive the change in spiral size would be more complicated, I think
Okay, imagine that you show that two consecutive sequences are equal, in that each pair of neighborhoods with constant time offset from each other are equal
Even if that happens, it could change once the spiral size increases.
 
Erik the Outgolfer
Erik the Outgolfer
yeah, that might be a problem, that's why I'm just making assumptions
although the time difference isn't a function of the spiral at all, it's just the number of steps between to go from step A to step B in the history
for some numberings, it may be impossible to prove if the piece will be moving forever, so yes, there should be a pattern
 
Deadcode
Deadcode
But of course it's a function of spiral size... I mean, imagine that you somehow have a pattern of time offset T_o that somehow survives for N repetitions... but then falls into chaos when the spiral size gets even larger.
So the pattern survived for T_o*N steps, but that didn't prove anything
 
Erik the Outgolfer
Erik the Outgolfer
well, in some cases, it might for the spiral system
for example, if the repetitive pattern of visited cells has a certain width, and it survives to a spiral layer that's way bigger than it, well...
 
Anonymous
@Deadcode The comments were only tangentially related to the challenge, and did not add anything to the challenge itself.
 
Deadcode
Deadcode
6:18 PM
@Mego Okay. Thanks.
 
Anonymous
However, they weren't outdated, and provided insight to something related to the challenge, so I opted to move them to chat rather than deleting them outright.
 
Deadcode
Deadcode
Right, I just think they're less likely to nerd-snipe somebody who might be interested, when separated by a link like that.
 
Erik the Outgolfer
Erik the Outgolfer
he's most likely referring to the purpose of a comment, and the comments technically didn't follow this purpose, so he moved them to chat
 
Deadcode
Deadcode
I would have been less perturbed by the discussion being cut off there (with any continuation being forced to be in chat) instead of all already-existing comments being moved.
 
Erik the Outgolfer
Erik the Outgolfer
hence the link to the chat room
 
Deadcode
Deadcode
6:23 PM
When a regular user moves a discussion to chat, the already-existing comments don't get moved. I think that would have been much better in this case.
 
Erik the Outgolfer
Erik the Outgolfer
I don't think mods can copy comments like that, and they should be preserved in the chat room, so...
 
Deadcode
Deadcode
I mean, the already-existing comments get copied, not moved. But here, they were moved, and that's what upset me.
I think nobody is going to visit that chat, without any snippet of discussion to draw them in.
 
Anonymous
@EriktheOutgolfer We can, but there's usually not much point in doing so.
 
Erik the Outgolfer
Erik the Outgolfer
yeah, no need to keep the comment section full of technically unnecessary comments, if you're going to move them then move them :P
 
Anonymous
When we move comments to chat, we have the option of keeping the comments on the post, but 99% of the time that's not useful
 
Deadcode
Deadcode
6:27 PM
And BTW, they're not 100% preserved in the chat room. The upvotes are erased.
 
Erik the Outgolfer
Erik the Outgolfer
the upvotes are of next to zero importance
 
Deadcode
Deadcode
Just saying.
Also, I hate that this chat system doesn't show timestamps for every line. And doesn't include seconds in the timestamps at all.
 
Anonymous
The only time it is useful, is if there are some comments that should stay on the post (in which case we'd have to not choose to delete when moving to chat, and cherry-pick-delete the comments manually). That doesn't happen often.
 
Deadcode
Deadcode
There isn't even any way to see extended timestamps on mouseover.
Can this be fixed?
 
Anonymous
@Deadcode This is true, because chat lacks upvotes
 
DJMcMayhem
DJMcMayhem
6:28 PM
It has stars
 
Erik the Outgolfer
Erik the Outgolfer
meh
if you're polling in the comments, maaaaaybe they're not exactly the right medium ;)
 
Anonymous
@Deadcode SE can change it, but most chat feature requests get status-ignored
 
Deadcode
Deadcode
The upvotes just show that somebody thought those comments were valuable enough to click an arrow. It takes a lot more to make somebody think a sentence is star-worthy.
 
Erik the Outgolfer
Erik the Outgolfer
chat doesn't have upvotes though
 
Deadcode
Deadcode
My point is just that moving the discussion to chat, instead of copying it, did not preserve the information.
And also that it makes it very unlikely for anyone to visit the chat link, because there's nothing to draw them in (as said above).
 
AdmBorkBork
AdmBorkBork
6:34 PM
@Deadcode re: Prove lack of trapping for any case -- Rook, Queen, King -- they simply follow the spiral? I also have a gut feeling that the Queen, Rook, and Bishop will never become trapped, though they may not visit every number like the Knight basically does.
 
DJMcMayhem
DJMcMayhem
Well, proving that the Bishop doesn't visit every number is pretty trivial :P
 
Erik the Outgolfer
Erik the Outgolfer
yes, those do in fact follow the spiral
 
Deadcode
Deadcode
@AdmBorkBork Mmmm, didn't even think of implementing queen moves. That could be interesting.
(And bishop and rook)
I was only thinking of finite moves
 
Erik the Outgolfer
Erik the Outgolfer
@AdmBorkBork um... I think the Knight does get trapped, 2016 IIRC
 
Deadcode
Deadcode
I mean of course it will turn out to be trivial if they only have those moves, but if they have those and others...
Well, there would have to be some restriction as to when they can use those moves
 
AdmBorkBork
AdmBorkBork
6:36 PM
@EriktheOutgolfer Yeah, but doesn't it visit all the numbers in between? Or does it skip a bunch?
 
Deadcode
Deadcode
Like, a minimum distance to apply a queen move
for anything shorter, it can only be a knight-style move
 
AdmBorkBork
AdmBorkBork
To be honest, I haven't watched the Numberphile video. No Tube of You at work, and I'm not caught up yet on my subscriptions list at home, so I just haven't seen it yet.
 
Deadcode
Deadcode
1,2 + 1,3 + 1,4 case is at Step 267244273664 and still chugging along. But there's such a huge "wake" between the lowest-numbered unvisited square and the highest-numbered visited square that trapping seems inevitable.
 
Erik the Outgolfer
Erik the Outgolfer
also, there's a bit of a problem regarding pieces such as the Rook, Bishop, Queen or King, there's an infinite number of possible moves
 
feersum
feersum
@Deadcode You could have abandoned squares without trapping.
 
Deadcode
Deadcode
6:42 PM
@EriktheOutgolfer It's not really a problem. You don't need to consider moves going outward beyond 1 step, because you're searching for the smallest-numbered square to go to.
(outward from center)
@feersum What do you mean, abandoned?
 
feersum
feersum
Ones that are never visited.
 
Deadcode
Deadcode
The wake is continuing to move forward.
No islands are being left behind permanently... yet at least.
I implemented pruning to be as CPU-efficient as possible, so in this particular case I can only say that position 161,623,115,056 it was 100% visited up to there.
For the next run, I'll want to make pruning happen more often, just to be able to see the size of the wake
 
Neil
Neil
6:57 PM
I would have tried 1,2 + 1,3 + 2,3 first, since |2,3| < |1,4|, hoping that would reduce the number of steps before trapping
 
Deadcode
Deadcode
@Neil Yes, I agree. But I dedicated 21 GB (foolishly) to this 1,2 + 1,3 + 1,4, so I can't now start up a concurrent 1,2 + 1,3 + 2,3 run now unless I give it much less memory
 
New Main Posts
New Main Posts
7:12 PM
3
Q: Ambiguous Locations on a Grid

BeefsterYou have a little robot with four distance sensors. It knows the layout of a room, but it has no sense of orientation other than being able to lock onto the grid orientation. You want to be able to find out where the robot is based on the readings, but it can be ambiguous because of the limited s...

code-golf grid
 
Deadcode
Deadcode
7:36 PM
@AdmBorkBork Here are some SVG path graphs then: path of the knight, path of the camel
 
AdmBorkBork
AdmBorkBork
Those are pretty cool
 
Adám
Adám
 
Dennis
Dennis
@Adám Done.
 
Adám
Adám
@Dennis Thank you Dennis, you're d'mod.
 
Deadcode
Deadcode
 
trichoplax
trichoplax
7:58 PM
Is there a 3d analog of the number spiral used for these?
 
Erik the Outgolfer
Erik the Outgolfer
the only 3D representation of a spiral I can think of is the tornado...
 
Deadcode
Deadcode
Something like a ball of yarn, maybe?
 
New Main Posts
New Main Posts
8:17 PM
0
Q: Find a Rocco number

vijayscodeI was asked this question in an interview but I was unable to figure out any solution. I don't know whether the question was right or not. I tried a lot but couldn't reach any solution. Honestly speaking, nothing came to my mind. A positive integer is a Rocco number if it can be represented eit...

programming-puzzle functional-programming equation
 
lirtosiast
lirtosiast
@trichoplax Sandboxed a challenge to create a 3D spiral
 
Deadcode
Deadcode
> Because all your computer's memory is taken up by the wildebeest simulation you're running, typing is very slow, so your code must be as short as possible.
LOL.
 
New Sandboxed Posts
New Sandboxed Posts
1
A: Sandbox for Proposed Challenges

lirtosiastGolf a 3D spiral Inspired by https://chat.stackexchange.com/transcript/message/48767657#48767657. The spiral used in The Path Of The Wildebeest is a contiguous mapping of the positive integers to lattice points in 2D. Your task is to generalize this to 3D. Specifically, create a function \$f\$...

 
Deadcode
Deadcode
On a serious note, it's just 1 thread. I can't think of any way to multithread a wildebeest type simulation.
But maybe we all have 8 threads with different parameters running ;)
 
lirtosiast
lirtosiast
8:33 PM
Has anyone created a leaper proven to fill the infinite chessboard that doesn't have any of the king's moves?
 
Deadcode
Deadcode
@lirtosiast Yes, a 2×king fills the board trivally simply.
All moves of a king but each one goes 2 times the distance.
 
lirtosiast
lirtosiast
You mean {(2,0),(2,1),(2,2)}?
 
Deadcode
Deadcode
No, not 2,1
 
lirtosiast
lirtosiast
But that only reaches even squares
 
Deadcode
Deadcode
Oops, you're right. It makes a very simple spiral but skips all even squares.
Here's the one you described. {(2,0),(2,1),(2,2)}
Fixed the {(2,0),(2,1),(2,2)} link.
It has a very interesting shape.
 
lirtosiast
lirtosiast
8:51 PM
Interesting
 
Deadcode
Deadcode
It is visiting every square, but it does seem that it could be possible it will eventually reach a higher level of turbulence and at some point become trapped.
@lirtosiast > The Chebyshev distance from the origin is an increasing function (The function wraps the origin in layers like a cubical ball of yarn)
I hope you don't mean strictly increasing. Because that's impossible, isn't it?
It has to hop over its own trail.
The challenge would be to do this without runaway distortions accumulating.
 
lirtosiast
lirtosiast
9:06 PM
Of course
 
Erik the Outgolfer
Erik the Outgolfer
@JonathanAllan btw, it looks like your recent wildebeest is a pretty hot topic rn... ;P
 
lirtosiast
lirtosiast
10:00 PM
@BassdropCumberwubwubwub I believe neither should be closed. The wildebeest should be kept as it was posted first, and the knight should be kept as it's a somewhat fundamental problem.
 
Deadcode
Deadcode
10:13 PM
I agree with @lirtosiast. Also, the fact that it gets trapped so soon offers a potential for golf opportunities (using very small arrays, etc.). And the more limited set of moves might have golf opportunities that don't apply to the more general case of the wildebeest.
And the fact that the trapped position was known ahead of time before the challenge was created means that it doesn't have to allow undefined behavior.
 
DJMcMayhem
DJMcMayhem
10:25 PM
CMC: Given two positive integers a and b, return the XOR of every integer in [a, b]
 
Deadcode
Deadcode
(Though as it happens, it does allow for undefined behavior anyway.)
 
New Main Posts
New Main Posts
0
Q: A Radiation Hardened Palindromic Quine

KrystosTheOverlordThere are radiation hardened quines, palindromic quines, but has anyone tried both?! I think not! RULES So the challenge here is to make a quine that if any symbol is deleted it prints its original source code, AND the quine needs to be the same read forward and backwards! THINGS TO AVOID Do...

code-golf quine palindrome radiation-hardening
 
Mr. Xcoder
Mr. Xcoder
@DJMcMayhem Test case please? I am not sure I understand what "the XOR of every integer" means
 
DJMcMayhem
DJMcMayhem
@Mr.Xcoder If a=3 and b=8, then you must output 3 ^ 4 ^ 5 ^ 6 ^ 7 ^ 8, which is 11
 
Mr. Xcoder
Mr. Xcoder
Ah, d'oh. Thanks.
 
Erik the Outgolfer
Erik the Outgolfer
10:32 PM
@DJMcMayhem Jelly, 3 bytes: r^/
 
Deadcode
Deadcode
10:43 PM
How about the O(log(n)) time version? :)
 
Mr. Xcoder
Mr. Xcoder
@DJMcMayhem Triangularity, 31 bytes (note: I had forgotten that Triangularity existed until I accidentally clicked on it on TIO)...
 
Erik the Outgolfer
Erik the Outgolfer
lol
 
Mr. Xcoder
Mr. Xcoder
...)...
..@IE..
.@)IEr.
"^"JE..
 
DJMcMayhem
DJMcMayhem
@Deadcode There's actually a O(1) version :)
 
Deadcode
Deadcode
er, I mean O(log(b))
The trivial version is O(b-a)
Which is in C++, int f(int a,int b,int n=0){return a>b?0:a^f(a+1,b);}
 
DJMcMayhem
DJMcMayhem
10:46 PM
Well, I guess a single XOR is technically log(n)... I guess you're right
 
Erik the Outgolfer
Erik the Outgolfer
um... yeah, this is when I feel two discussions might've interfered with each other :P
 
Deadcode
Deadcode
O(b-a) - technically O((b-a)*log(b))
 
DJMcMayhem
DJMcMayhem
Now if only I knew how to do radiation hardening
 
Erik the Outgolfer
Erik the Outgolfer
in V? good luck... :P
 
Veskah
Veskah
Clearly just double everything up in a comment :^)
(For real: Radiation hardened code is straight witchcraft most times)
 
Erik the Outgolfer
Erik the Outgolfer
10:55 PM
> — Radiation hardening: Making your program retain its behavior even after the removal of any one of its bytes.
— Alright, so I decided that this is the one that gets removed.
 
Deadcode
Deadcode
path-of-the-wildebeest-where-trapped.svg - I'm surprised how close to the surface she was when she got trapped.
 
Veskah
Veskah
@Deadcode That pattern duplicated would make a nice wallpaper
 
Deadcode
Deadcode
@Veskah Including the dot where she gets trapped, or just a portion of the inside with no edge visible?
 
Veskah
Veskah
Either, really
 

« first day (2922 days earlier)      last day (1904 days later) »