The Nineteenth Byte

randomra

08:00

added to standard loopholes:

0

A: Loopholes that are forbidden by default

Changing your username to fit a username-dependent challenge Changing your username or registering a new one to make your score better in username-dependent challenges.

xnor

i approve of that, though I don't think we should have username-dependent challenges in the first

randomra

red vs blue is ID dependent too

xnor

oh, right

randomra

it's good for classification

xnor

that sounds like a legit use case

does your ID change with username?

randomra

08:02

and there is the gravatar one

xnor

which one is that?

Sp3000

I wish there was a VTC for "too gimmicky" sometimes...

randomra

don't think so, but you can register a new user

xnor

maybe we need a "Challenges that people keep posting that have been way overdone"?

for new users in particular, there seems to be an irresistible appeal of "print X without X"

Katenkyo

@xnor No, because someone could find an original way of using them.

xnor

08:04

or "do X without Y"

@Katenkyo you're right, that's a snippy title

New Loophole Proposal

2

A: Loopholes that are forbidden by default

Changing your username to fit a username dependent challenge Changing your username or registering a new one to make your score better in username or userID dependent challenges.

randomra

13

A: Are Quines played-out?

No, quines are not played-out I believe there will always be new ways of formulating good challenges, and some of those yet to be made good challenges will involve quines. Saying "we've asked every quine problem there could ever be" is like saying we've thought of every math problem there could ...

Katenkyo

If you use such a title, no one will ever try to re-use them, which would be pretty bad for this site

xnor

what I really mean is "challenges for which many similar variants have been done so you should make sure yours adds something particularly novel"

Katenkyo

@xnor Much better :D

randomra

08:06

this was discussed with quines in particular

Katenkyo

I agree that quines are often pretty much the same thing

randomra

@xnor that's better

Katenkyo

The problem is the lack of reflexion behind questions. Those type of challenge aren't hard to write, but people underestimate how hard it is to write a good and original one.

xnor

indeed

and they often are interesting, just new users don't have the context to know that many others found them interesting in the past and posted lots of similar things

also, the "X without Y" thing often fails with esoteric languages that a user who thinks in terms of c/python/java wouldn't anticipate

in that esoteric languages would not have used Y anyway

or it doesn't make sense to define Y for that language

Sp3000

Well to prove my point I've posted why I don't like this challenge :/

Katenkyo

08:11

@xnor I saw that, I've read like50% challenge of this site before signing in, and often, when a Q was closed as dupe, it was created by a new user.

Sp3000

Print X without Y is one thing, tying usernames is just worse

Katenkyo

@xnor Those type of challenge are hard to define, really hard.

xnor

@Sp3000 why does your method no work for consecutive chars?

Katenkyo

If you want to annoy esolangs, "without []()" could be nice, but bash and some other language don't use them at all etc

Katya

It's not very good honestly. For instance I could trivially beat Sp3000's entry with

"J`sx`":)

Sp3000

08:13

@xnor In your case it would be "wmnq":), but since your username has n you can't do that

(you could map by some other increment other than +1, but that's another byte)

xnor

oh, i see

but your username is 6 letters, so I could still beat that :-)

Sp3000

Yeah, but Katya could tie :P

xnor

there's grc...

Sp3000

...

I think we have a winner

xnor

3 chars is username min

there are people whose usernames are solely numbers, so that's allowed

if this user became active: codegolf.stackexchange.com/users/6983/888

Katya

08:18

people with unicode characters in their name are probably annoyed

Sp3000

37 24* (5 bytes)

xnor

double 444 in Pyth

Sp3000

I mean 37*24

Ahaha or that

xnor

Pyth reads multidigit numbers, right?

Sp3000

Yeah, y444

xnor

08:19

apparently it does

oh, apparently you subtract the length of your name

so now you want a really long name you can compress

Sp3000

I missed that part too

xnor

I've got /y^T8 9 for the user 22222222

for a score of -1

ooh, there's user 1001010

Sp3000

74Yb

xnor

.B74

Sp3000

I think the other username is better, '28*

xnor

08:33

that has a 2 :-)

Sp3000

... '1)8*

You saw nothing.

xnor

a bold move, posting a new version rather than editing

Sp3000

Oh it's the same then

Hahaha well mobile

xnor

.B937 for 1110101001

gives -5

you can probably do better with ss-3-1415926535897932384626433 if you have pi built in

though the prefix is unfortunate

Sp3000

That's a lot of digits though, more than a float

xnor

08:43

user 1000000000 is ^T9

Sp3000

How are you finding these?

xnor

i just went to users and let 000 autocomplete

might not work on mobile

Sp3000

I'm done with food, so I'm on my laptop now :)

xnor

if you know how to query these, it's probably way better than what i'm doing

Sp3000

@xnor data.stackexchange.com/codegolf/query/322396/numbers

4,(;5* for 123123123123123

xnor

08:50

18446744073709551615 is 2**64-1

Sp3000

Y'@i#(

Actually...

GG#(

xnor

whoa, what's that?

Sp3000

16^16-1

xnor

oh, nice

that's going to be hard to beat for the fantasy user challenge

Sp3000

Indeed

xnor

08:54

actually, what if you look at the longest-named users?

Sp3000

That's what I was thinking...

xnor

if the char limit is high enough, anything compressible should win

Sp3000

Char limit is 40 I think

Hmm the query takes a while because the output is long

But in any case it's

SELECT *
FROM Users
ORDER BY LEN(DisplayName) DESC

Nothing longer than 18446744073709551615 looks compressible to me

xnor

well, we could fantasize some more

what if you could choose any valid username?

Sp3000

(except maybe that pi one)

What constitutes a valid username? Any char?

xnor

09:00

"Display Name can only contain letters, digits, spaces, apostrophes or hyphens and must start with a letter or digit"

Sp3000

Is there a length limit?

(Based on the data query I'm guessing 40 but I'm not sure)

xnor

the entry field lets me put in only 30

and that's the length of the pi username

so unless your query gives longer ones, that's probably it

Sp3000

Well the longest actual usernames lists are 30, but the query field seems to allow 40

Let's go with 30 though

xnor

by coincidence, 2^99 has 30 chars

and no 9's

Optimizer

@xnor that's what happens to new questions also (if a challenge is based on question names). They are not allowed.

xnor

09:07

sorry, what's that in response to?

Optimizer

"new users wont be able to participate"

xnor

is there a source for that?

perhaps worth adding to the loophole

Optimizer

I have seen multiple questions in past

one was by martin only

xnor

@Sp3000 there's of course t^T30 to do 30 9's

i don't know

you have an overlapping 8 though

Sp3000

whistles

I keep switching between CJam and other things and thinking I get numbers for free

e.g. KK# is unfortunately 27 long

xnor

09:15

how high do the numbers go?

Sp3000

Up to 20

Optimizer

what are you working on ?

Sp3000

I think usernames can have dots so PP would also be a prime candidate, but that's 34 chars

xnor

any chance squaring is a single char?

Sp3000

(we're seeing how bad exactly the username challenge idea is by trying to find the worst case by making up usernames)

Nope, not that I know of

xnor

09:17

what about doubling?

Sp3000

Only Pyth does that :P

xnor

i've been spoiled

Sp3000

The only languages I know with a single char for squaring are Deadfish variants... not very useful here

xnor

i found how make a 63-char number output in 2 chars of Pyth

not useful here, but i think it's cute

Sp3000

If you include languages which give some sort of message (to STDOUT for some reason) on error, things could get a little crazy

isaacg

09:20

@xnor What is it?

xnor

CG

isaacg

Nice.

Of course, yG is the real winner

Around 1 billion characters

xnor

haha

Sp3000

Should have said that last line before I plugged it in online :P

xnor

i was wondering why i wasn't getting output :)

isaacg

09:22

I tried it on my machine, it ate all the ram. Almost crashed.

Sp3000

"An error occurred in the application and your page could not be served. Please try again in a few moments."

hifives xnor

Katenkyo

@Sp3000 Are you trying to get some negative scores because of long username/short arithmetic code? ^^

xnor

yup

Sp3000

We already have negative scores, we're trying to get negativer

Katenkyo

How negative are you actually? -10?

xnor

09:24

with existing usernames, -16

with fake ones, -25 i think

Katenkyo

Oo

Sp3000

-26 actually

Katenkyo

Wow, cracked !

xnor

@Sp3000 i missed your 4-char then

Katenkyo

Well done guys !

Sp3000

09:25

That's cos I didn't post - Y99#, CJam doesn't need the space

Optimizer

is there an existing user name like that ?

xnor

@Optimizer nope, but there one that's the number 2**64-1

Sp3000

Hence the "With existing, -16. With fake, -26"

Optimizer

@xnor link ?

Sp3000

Damnit xnor

xnor

09:27

hahaha

VisualMelon

nominate Sp3000

Sp3000

You post again, I'll wait

xnor

no, you were first

i do insist

Sp3000

18446744073709551615

111 2

isaacg

I'm trying to do that challenge in Pyth, and it's really hard. G and C are the primary ways to generate strings in Pyth.

Sp3000

09:31

Unrelated, but sometimes I wish the Unscramble the Code challenge was still going. It'd be interesting to give puzzles which test how well you know the syntax of a language...

xnor

@isaacg man, you have bad luck with your letter choices

i vaguely remember some other challenge being ruined by G

Sp3000

Has anyone even written a Pyth quine yet?

xnor

1?

Sp3000

Ignoring that sort of thing

xnor

the empty string got banned :-)

isaacg

09:32

jN*2]"jN*2

That's the shortest one I know.

xnor

err, did we destroy your server?

Sp3000

That's missing the ] I think

Oh, wait the ] is a mistake?

Er...

isaacg

jN*2]"jN*2]

I meant

It's needed

Sp3000

Ah right

Katenkyo

@Sp3000 I think these type of challenges could be relaunched sometimes, for the glory of fun :)

With some modification, for sure

Sp3000

09:36

Ahaha, CnRs do feel like they could be repeatable sometimes...

Simple example for Python: Unscramble ()-...25::>def or so it produces no errors

xnor

interesting

just making it valid code

Katenkyo

@Sp3000 I think it could be, just ban cops answers from the previous one.

Sp3000

(I put the space between f and o because otherwise it wouldn't show obviously)

trichoplax

RoboCritters is just waiting for a quine question

(or a quine answer somewhere...)

xnor

@Sp3000 maybe this is simple, but i'm stuck

i have no idea how to get rid of two colons

Sp3000

09:46

:D

isaacg

I'm working on it too

I can only use the colon on the def, and nothing else.

xnor

i can't see a way to use it as a slice without []

isaacg

@Sp3000 Python 2 or 3?

Sp3000

Woops should have specified, Python 3

isaacg

3.5?

Sp3000

09:53

Nope, just testing in 3.4

Might also work before that too, will check

Yeah it should work fine in all of Python 3

isaacg

Alright, got the two :

Now I'm working on the .

isaacg

10:12

Got it

def o(r:2)->5:...

Sp3000

Nice! :)

isaacg

That was fun

I have no idea what that means

->? o(r:2)?

Sp3000

The first is function annotations, which is mainly for def square(n: int) -> int: return n**2

And the ... is the Ellipsis object, which is more for numpy and such

isaacg

I know ..., but I'd never seen the annotations before.

Sp3000

Well now you do! Useless in golf, but good for code documentation I guess :)

isaacg

10:16

MIght be helpful in Pyth, I guess.

Probably not

Sp3000

Well the annotation shows up in the REPL, so it might if you use that for debugging

But docstrings might be more useful

randomra

@flawr may I use your cat&catcher controller as a starting point for my timetravel koth?

Katenkyo

11:55

The Great Wumpus Hunt KotH is so much random, it's sad :/

New Main Posts

12:09

0

Q: Chaining autograms

Write a program that takes one integer input n and one format letter and outputs n autograms in the format This sentence contains ? "A"s, ? "B"s ... and ? "Z"s. Each question mark ? represents the number of times the the following letter appears in the text The integer (let's call it n) is used...

code-golf kolmogorov-complexity

randomra

@Katenkyo oh, I thought that's an older challenge, I almost haven't noticed it

Katenkyo

It's a new one, but you often spawn on the wumpus, changing completely the outcome :/

Katya

that seems like a rather large oversight

The E

Yeah it's slightly annoying :)

Katenkyo

Total rounds: 1500
Humans killed by arrows: 4
Wumpus victories: 923

Can't lower him below 60% win >_<

Ho, 59% win for the wumpus, better now :D

Total rounds: 10000
Humans killed by arrows: 1
Wumpus victories: 4185
Stats fort only 4x my bot, so i'm losing cause of others... ok, I take note

New Main Posts

12:28

1

Q: Extending OEIS: Counting Diamond Tilings

I promise, this will be my last challenge about diamong tilings (for a while, anyway). On the bright side, this challenge doesn't have anything to do with ASCII art, and is not a code golf either, so this is actually completely different. So just as a reminder, every hexagon can be titled with t...

fastest-code combinatorics tiling

Martin Büttner

12:54

hm, test cases for Boggle Board Compression ... what do people think about just having N-2 strings from length 3 to N, and the test cases are then just all prefixes of that (i.e. just the first string, just the first two strings, just the first three strings, etc. up to all N-2 strings).

randomra

that would have less value than a different set for all cases

if one algorithm is good on only that set, it might have an unfair advantage

but the new, extra strings are longer, so that might mitigate this problem

Martin Büttner

I'd probably just make the strings random

randomra

how big alphabet would you use?

Martin Büttner

I'm not sure yet. small enough that there is an interesting amount of overlap

also depends on N

randomra

yeah, I have difficulty imagining it with 32+ sized alphabet

Katenkyo

13:07

+1 randomra, I think between 12 and 17 would be more than enough

You could go up to N=30 with lots of overlappings/interesting things

Martin Büttner

I might try just using the alphabet and see how much overlap I get from random strings

Katenkyo

I think hexadecimal would be more interesting, the alphabet might be way too large :x

Have to be tested

Martin Büttner

yeah I'll check it out

Katenkyo

How much action MAXIMUM do you think I could leave as choices in a KotH?

randomra

my (completely random) guess would be around |alph|=16-20, N=32-50

Lembik

13:23

hi all

I was thinking posting a chess challenge..

how many queens does it take to cover an n by n chess board

does anyone know if this has been done?

Martin Büttner

1

Q: Domination problem in chess

Given an mxm chess board, find a minimum number of pieces of the given kind and place them on a chess board in such a way that all squares are either occupied or attacked by at least one piece. Input The input consists of two parameters: m, the number of squares on each side; and a letter which...

code-golf chess

Lembik

interesting!

Katenkyo

It is a common problem in computing world :3

Thomas Ross

https://twitter.com/dabeaz/status/604238628437377024

PSA: The "0x" prefix on hexadecimal numbers is pronounced "hug-kiss". As in "hug-kiss dead beef". Use it in your next talk.

Lembik

mine would be fastest code and the winner would be the person who got to the largest n

Katenkyo

13:26

Mainly known in the form "how to put n queens on a n*n chessboard"

Lembik

@Katenkyo isn't that usually so they don't attack each other?

@Katenkyo is there a known solution for my version of the problem?

Martin Büttner

@ThomasRoss twitter links onebox if you post just the link

Katenkyo

I think the solution for yours is derivated from the solution of the n queens for n*n

Lembik

ajc.maths.uq.edu.au/pdf/15/ocr-ajc-v15-p145.pdf

Thomas Ross

PSA: The "0x" prefix on hexadecimal numbers is pronounced "hug-kiss". As in "hug-kiss dead beef". Use it in your next talk.

Tweeted by dabeaz on May 29, 2015 at 10:51 AM

TIL

Lembik

13:28

@Katenkyo It's NP-hard apparently

Katenkyo

@Lembik I don't know how NP it is, not familiar with this term ^^'

Lembik

:)

Katenkyo

Never looked at that when working on an algorythm ^^'

More concerned about suppressing the worst solution it will produce, than if it can find the best one

JNV

Hi, I have a quick question. How long should I wait for responses after posting something in the sandbox? Is one day enough?

Lembik

@MartinBüttner I really love your "Extending OEIS" question

@JNV two days is good

Katenkyo

13:34

@TheE Thanks for the edit on wumpus, couldn't do it by myself

@JNV Depends of the time you need to make it perfect

Martin Büttner

@Lembik Maybe that'll get you to answer a challenge for once? ;)

Lembik

@MartinBüttner :) I can only take your criticism on the chin

@MartinBüttner we should have a whole series of them

JNV

@Lembik Thanks. @Katenkyo I think it's pretty fine now. But I'll wait a bit more. Thanks.

Lembik

OEIS challenges that is.. not criticisms of me :)

Martin Büttner

Sure, but that depends on finding sequences that are a) interesting b) conceptually simple and c) haven't been studied thoroughly yet. I think the diamond tilings was a really lucky find. I don't expect to find more of these easily.

Vioz-

13:37

Hey y'all :)

Katenkyo

@Vioz- Hi

Lembik

@MartinBüttner agreed

Katenkyo

It could be the second time PPCG contribute to OEIS, would be nice ^^

Lembik

it would be great!

I feel there is a lot of talent in PPCG that should be contributing more than CJam to the world :)

Martin Büttner

@Katenkyo that reminds me, I still need to propose the fix for the Fission numbers

(that's not an important contribution though, those were merely typos)

Katenkyo

13:43

@MartinBüttner So it could be the third one ^^'. Even typos are important to correct if someone wanna work with it someday

Guys, i've a problem with the koth i'm designing

Currently, players have 5 "actions" possible + 1 between 5 capacities that can be used. Way to much isn't it? :/

PhiNotPi

I'm not sure what you mean by "1 between 5 capacities"

trichoplax

Why is it too much? Too complicated to write answers for? Too many possibilities for a decision tree? Too unwieldy for I/O?

PhiNotPi

have you posted it in the sandbox yet?

trichoplax

Often simplicity makes for a better challenge and more answers, but if there's a good reason for adding a feature then add it

PhiNotPi

5 actions really isn't too much. Rock-Paper-Scissors-Lizard-Spock had five actions.

trichoplax

13:48

Is that "a choice from 5 options" or "5 choices to make, each of which has several available options"?

@PhiNotPi Agreed - 5 options to choose from is definitely fine. And 8 directions to choose from in a grid based KotH works well too

Katenkyo

So basically, you dirige 6 cells, which can : Clone,Move,attack,explode. Each of them can have 1 (special case of 2) capacity between : Eat,Fire,Thunder,Merge,Heal. It's about guiding and expending/merging your cells to fight your opponent

Each cells have stats that are set by the player at start, and change with merging/eat

@trichoplax I agree with that point, that's why i'm scared it's a lot more to complicated

@PhiNotPi Still redacting it

(it's on a 2D grid, 8directions)

trichoplax

That doesn't seem obviously too much. It might turn out to be with testing, but I'd run a few basic strategies and see

Does a player choose which capacity to give each of its cells? Or are the capacities assigned by the KotH controller?

Katenkyo

@trichoplax Ok, so i'll do that. Finishing the redaction, then will draw some basic strats

Tricho -> It's the palyer's choice

player*

I think it is more interesting if the starting setup can be modified

trichoplax

Did you know you can edit by pressing up, or by clicking on the down arrow if you hover left of the post?

Katenkyo

@trichoplax No, didn't, thanks !

trichoplax

13:57

@Katenkyo I think so too. It also means that early answers can focus on a specific capacity and give it to all their cells, which allows simple strategies early on

Katenkyo

Time for a pause, brb

@trichoplax Yeah, but if it is complicated in itself, people won't bother trying it ^^'

PhiNotPi

14:14

Ready for postage? meta.codegolf.stackexchange.com/a/5355/2867

trichoplax

@Katenkyo Good point. If you want it to be clear that a strategy doesn't need to use everything mentioned in the question, you can include a simple example answer that only uses one or two features, then people can work to beat that

Martin Büttner

@PhiNotPi does it maybe need some definition of how input/output can be encoded for the proof of Turing completeness, since actual I/O should not be considered?

PhiNotPi

What would that definition be?

Is I/O even needed for Turing completeness?

Martin Büttner

well you need some input to your computable function and the result of said computable function needs to go somewhere, right?

it's usually sufficient if that is some internal state.

trichoplax

@PhiNotPi I see some unmatched asterisks

(no footnotes)

PhiNotPi

14:19

@trichoplax I know.

Katenkyo

@trichoplax I think 4 bots would be a reasonable amount, allowing to some interesting possibilities that doesn't have to use every features

PhiNotPi

For example, instead of writing a program that takes a number as input and outputs whether or not it is prime, like $input=<>; primeTestHere you could simply write a new program for each number you wanted to test, like $input=2; primeTestHere then $input=3; primeTestHere then $input=4; primeTestHere etc.

trichoplax

@Katenkyo Does "bots" here mean cell types, total number of starting cells, or answers to the question?

PhiNotPi

Painfully inconvenient, but that's not the point.

Martin Büttner

@PhiNotPi of course, that's how I expect it will go

but that doesn't even matter for the actual code

my concern is more for people who have to prove Turing completeness

trichoplax

14:25

@Katenkyo Am I right in thinking that if you start with 4 cells, each will have a certain capacity which cannot be changed, and all the descendents of a given cell will have the same capacity as it?

PhiNotPi

If this were your challenge, how would you define it?

Martin Büttner

but "When proving Turing-completeness, you may assume that input is hardcoded into the program, and output can be read off the program's internal state." might be sufficient

I wish Peter was around :/

PhiNotPi

That's probably good.

For example, would halt iff true but infinite loop iff false count as output?

Martin Büttner

ehm... I'm not sure about that

PhiNotPi

I think it would, since there would also be a program running at the same time that is infinite loop iff true, halt iff false.

Katenkyo

14:27

@trichoplax A bot means an answers. A bot will control all of its cells each turn (there's an extra parameter for each cell that you can use like you want, it could be easier to remind the role of each cells, and act accordingly)

Martin Büttner

@PhiNotPi not necessarily

trichoplax

@Katenkyo Ah I see. So you're suggesting writing 4 example answers?

Martin Büttner

isn't that the crux of the halting problem?

you can determine if a problem halts, but you can't determine if it doesn't.

Katenkyo

@trichoplax Yes, you are right to assume that, by cloning, I really mean cloning ^^

@trichoplax That's it

trichoplax

That sounds good. So the example answers could each focus on just one capacity, and then people who add later answers can start to combine different capacities that support each other and build up more complex strategies

PhiNotPi

14:29

In the meantime, I'll just copy-pasta what you said.

trichoplax

@MartinBüttner I wish Peter were around (in case that helps with missing him...)

PhiNotPi

@Peter, if you don't show up in the next five minutes, we will rename this site to "Programming Popularity Contests and Code Bowling".

9

Maybe that will get him?

Katenkyo

PPCCB, nice name ^^

PhiNotPi

14:46

What should I title my post?

"Output every hating program (write a parallel interpreter)"

3

Or should I drop the stuff in parentheses?

trichoplax

(just had to wait for the 2 minute edit window to expire before starring the amusing typo ;) )

PhiNotPi

"If your program outputs itself, it is probably wrong."

probably going to add that as a rule

Transcript for