The Nineteenth Byte

@lembik yes I noticed the 5th comment vote so I'll definitely be adding visualization now. I'm considering filling a folder with an image for each arrangement but that would be bad if they choose a large n. Maybe I'll display the number of arrangements first and then say "Are you sure you want to create this many separate image files?"

Martin Büttner

9:02 PM

@Optimizer eval'd would be weird, if the input doesn't eval to a single value

Optimizer

it just contains all the values

Martin Büttner

(or if the input contains code that doesn't terminate)

Optimizer

that would be eww even in current situation

Martin Büttner

in the current situation you wouldn't write a program where that could happen

if you added that as an interpreter feature, you literally couldn't take input that is code that doesn't halt

Optimizer

the variable can be lazy initializing

I dont see why this needs convincing ..

so many places we try to use ea just because we want to reuse the input

Martin Büttner

9:06 PM

ea doesn't evaluate straight away either. but lazy initialisation is a good idea (although it would be quite a special case then...)

Optimizer

even a non eval-ing input variable would be so handy

Martin Büttner

true

Optimizer

but eval'd is just a killer

aditsu

9:36 PM

finally, pushed the code and closed ticket 2 :) feel free to pull from hg, build and test

Martin Büttner

nice

is the online tester updated yet or will that have to wait until the next release?

aditsu

next release

Martin Büttner

kk

Lembik

@trichoplax my problem is that as there was only one answer, I am not sure it is worth posting my follow up questions

I thought it was a nice problem!

Martin Büttner

@Lembik it really is!

Lembik

9:49 PM

@MartinBüttner thank you! Do you think I should just leave it a week or two

and see if people pick it up?

the associated article has some very smart ideas people might be able to use

I mean the one that solves the problem without worrying about the collisions during rotations

feersum

that one of the problem restrictions isn't relevant until large N is a bit non-nice to me

I could re-post my answer from here and get the highest score

ok, it would have to be modified not to connect the end

Lembik

@feersum how would you know which values it is correct up to?

@feersum we don't know the smallest n for which collisions during rotations make a difference

feersum

it would be your problem :P

Lembik

at least not yet

that's not the sort of thing we want to see here! :)

actually I asked on math.se if someone could prove the smallest n for which it makes a difference. I can see it does for n = 25

but maybe less as well

I could have asked on puzzling.se .. I would have got a lot of answers immediately

then when they proved wrong or unsupported a lot of close votes on the grounds the question was impossible :)

(sorry puzzling.se, you only have yourself to blame)

feersum

n=23 is over 2 billion for the easier version, so it is probably impossible to calculate enough of them to even reach the restriction in 1 minute

Lembik

9:56 PM

@feersum I am not sure.. did you look at americanscientist.org/issues/pub/how-to-avoid-yourself ?

feersum

no

Lembik

americanscientist.org/issues/pub/how-to-avoid-yourself/4 in particular

computers weren't very fast in 1972

I imagine one minute on my computer was more computing power than they had in a lifetime on one of their computers

feersum

if you actually wanted to consider rotations, you would have to use a much less efficient and more complicated method

Lembik

@feersum you mean to worry about collisions during rotations?

do you mean my problem is much harder than their's?

feersum

naturally

Lembik

10:00 PM

ok.. well I can't say :)

as I don't even know what the more sophisticated methods of A. J. Guttmann were yet

feersum

it has all the restrictions of the other problem plus some kind of pathfinding

Lembik

It's not clear to me that you actually have to count each shape in turn

it seems you can get a lot in one go potentially

some sort of divide and conquer I have in mind

feersum

it could be doable, but seems like too much effort for PPCG

Lembik

or maybe something cleverer

@feersum maybe... :) I mean some of the answers here are awesome and need a lot of brain power

@feersum someone might want to spend an evening working on something

people go to a lot of effort for no reward

see math.stackexchange.com/questions/562694/… as one of thousands of examples

I don't see why people would go to less effort on PPCG than they might on any other .SE site

@feersum unless you just mean programmers are lazy :)

feersum

it could happen but there are lots of other questions too

Lembik

10:07 PM

sure but there are always lots of other questions!

that's true everywhere

the question is, are they more interesting to everyone :)

randomra

@Lembik what is your question? i can't sem to find it

Lembik

10

Q: The number of reachable snake orientations

This challenge is not about the game Snake. Imagine a 2d snake formed by drawing a horizontal line of length n. At integer points along its body, this snake can rotate its body by 90 degree. If we define the front of the snake to be on the far left to start with, the rotation will move the back...

In this case I feel a good answer would get a new oeis entry :)

something to be proud of

@randomra I have two questions going.. the other one is about counting kmers

the second question may be about to be cracked by FUZxxl and the magic of suffix arrays ...

randomra

ah, I thought you had some new hard problem

Lembik

@randomra well it is about one week old and hard :)

@randomra do you mean my challenges are too easy? :)

randomra

I like these and think about them although usually you only see an upvote...

Lembik

10:15 PM

thanks!

Sp3000

The day when a Lembik problem takes 10 minutes to solve is the day when PPCG graduates

Pyraminx

hey golfers

Sp3000

Hi!

Doorknob

@MartinBüttner a.) Sorry, nope. b.) 531 results... O_o

Perhaps I'll write a small script to un-CW all the [tips] questions and answers, because 531 is a bit much

Martin Büttner

@Doorknob huh, is there no way to un-CW all question with all its answers?

randomra

10:21 PM

@Lembik if collision would only count at turning start&end positiions do you think that would be oeis.org/A037245?

your example undoable shape can be made in that case

Doorknob

16

Q: Allow "Remove Wiki" on a question to cascade down to answers

We recently had a question appear on the "Hot Network Questions" list, and it attracted a lot of attention to our site. We are very grateful for the views, and the influx of new users (many of whom appear interested in sticking around, which is just fantastic for a beta site that needs more part...

feature-request moderator-tools community-wiki

Martin Büttner

meh :/

rolfl

Hey @Doorknob... educate me ... ;-)

I have judged this question on Code Review as off-topic both on CR and PPCG. Still, I thought I may be wrong, and sometimes things can be 'adjusted' to make it work. Thoughts?

Martin Büttner

@rolfl if he actually wants to cut it down to minimum byte size it would be on topic as a tips question, but if he just wants a short-but-readable version, it's not. however, even in the former case, I'd probably CV as duplicate on our standard quine challenge, because I'm sure it has the shortest known JS quine.

rolfl

There is such a thing as a short-but-readable quine ? ;-)

actually, the longer it gets, the less readable it is.

I think you're right, and it's too ambiguous.

aditsu

10:34 PM

there are short and readable quines, but it depends on the language :)

www0.us.ioccc.org/1994/smr.hint

feersum

Do we have a decision problem quine challenge?

That is, determine whether the input is your source code or not

Martin Büttner

@rolfl It seems like you've been rule lawyering the guy a bit though with this comment ;)

feersum

I read about this recently and it seems like a much better problem than standard quine, since it disallows degenerate answers

Martin Büttner

unless "Questions about code whose functionality depends on the source itself are off-topic" is CR policy ;)

@feersum yes

aditsu

@feersum codegolf.stackexchange.com/questions/11370/…

rolfl

10:40 PM

Yeah, it is rule-lawyering... I know that.

I wasn't trying to make it subtle ;-)

A quine is a particularly hard thing to review, because you cannot suggest any valid changes without changing the output.

and if people on Code Review want to play code games, they should do it somewhere else. Not that we can't have fun... though. We like competitions. Just ask PCG.

Nathan Merrill

would it be possible to produce a quine that would still work with a single character deletion anywhere in the code?

aditsu

QQ ? :)

feersum

degenerate echo quines work, for starters

like PHP without <?, etc.

or do you mean the original unmodified source?

Nathan Merrill

the person writing the code doesn't get to decide which character is removed

basically, a quine that is still a quine with any 1 character removed

rolfl

That would be unlikely.

feersum

10:45 PM

then my above comment is still valid

Martin Büttner

@aditsu not itself a quine though

Nathan Merrill

example feersum?

Sp3000

That can be easily solved with the Befunge technique mixed with the 2D quine

feersum

ok, my PHP quine is "abc"

rolfl

For example, most languages require whitespace somewhere....

Martin Büttner

10:45 PM

@rolfl you don't know the kind of languages we're using :D

aditsu

@MartinBüttner does it actually do Q twice? I'm not sure it's clearly defined

Martin Büttner

I'm pretty sure it does

but I'm not an expert on HQ9+ either :P

aditsu

you may be right

@rolfl whitespace is the easiest problem to solve

just double them :)

Martin Büttner

I knew I had heard it somewhere... I thought it was a PPCG challenge but no

@NathanMerrill github.com/mame/radiation-hardened-quine

Nathan Merrill

cool

Martin Büttner

10:53 PM

@aditsu it's not that trivial, because there might be whitespace that doesn't automatically get replicated exactly in the quine

(because you manually replicated it in a string)

aditsu

ah

Sp3000

I like the issue on that github post - is it meant to quine the original or the one deleted

aditsu

well, depends if it's a quine that really constructs itself rather than taking an easier approach :)

Martin Büttner

oh okay

so it's not what Nathan asked for

I'm still surprised this hasn't been asked yet

Sp3000

I was just asking - because when I said it was easy in Befunge I thought Nathan meant outputting the original program

Martin Büttner

10:57 PM

although it does seem impossible without reading your source code, because even if you manage to get the framework redundant, there's generally no way to notice if a character has gone missing outside the actual string construction

this reminds me... did anyone ever write a quine in marbelous?

doesn't look like it

Sp3000

Maybe if you had a language where 11 pushed two ones and deleting either outputted 1 automatically with no newline

Martin Büttner

well yeah, in CJam 1 and 11 are both quines

but I wouldn't count that

in /// everything not containing / is a quine

feersum

same deal as PHP

aditsu

related: codegolf.stackexchange.com/questions/11383/…

Sp3000

^^ that's the one I was thinking of

aditsu

11:05 PM

that's a fault-tolerant way to point to my comment in befunge :)

New Sandboxed Posts

11:19 PM

0

A: Sandbox for Proposed Challenges

Missile War Background So here's the situation, we are at war! The last line of defense between them and us is the two walls defining us: +.....................................................+ +.....................................................+ +..............................................

Martin Büttner

11:34 PM

@PeterTaylor re normal distribution: someone else also requested a ruling on how accurate the results have to be. I tried fixing that by stating that the chosen algorithm must by exact in theory (ignoring datatype and PRNG limitations) and that all arithmetic needs to make use of the full width. let me know if you think that has similar flaws as my phrasing for the shuffling challenge.

Sp3000

Does exact in theory mean that it needs to be able to output ridiculously large/ridiculously small numbers with nonzero probability?

(i.e. numbers probably outside the float range)

Martin Büttner

no, it means that given a perfectly uniform source (of reals, not floating point numbers or something), it should produce a perfect normal distribution as output

feersum

should it be able to put out irrational numbers? :P

Martin Büttner

the algorithm?

absolutely

the implementation doesn't need to though :P

Peter Taylor

@MartinBüttner No, I think it's ok. The fact that the question necessarily requires the use of non-integral numbers makes a significant difference to its natural interpretation.

feersum

11:41 PM

if a language had only a random integer generator then it would be impossible to use?

Martin Büttner

@feersum you could draw integers in 0...MAX_INT and rescale that into the required range

that's a good point though

I should specify the minimum range the uniform distribution should be taken over

how is that? (just edited)

feersum

or you could assume you have a function for generate numbers from 0 to 1 which does not count as part of the code

Martin Büttner

hm, interesting, but I'd rather see people work with what their language gives them

feersum

in other words, exclude languages w/o random float generator

Martin Büttner

@aditsu wow, that feature request was handled quickly :D

aditsu

11:51 PM

well, it was simple and not controversial, so while I was at it... :)

Martin Büttner

thanks :)

hm, maybe I should include one instalment in the randomness series to actually write a PRNG?

for example, it looks like we never golfed a Mersenne Twister (although I can't tell how much fun that would be)

feersum

x << 23 | y >> 57 | x & ~9039236237LL ... doesn't seem that fun

Sp3000

Well there's been this

Martin Büttner

@feersum huh, I thought it would be trickier

@Sp3000 huh, interesting... I think that's quite different though

Sp3000

Well if you're doing specifically Mersenne Twister, then yeah it is - just thought I'd point out something related :)

Martin Büttner

11:57 PM

yep, thanks

I mean, maybe there's a more interesting one out there

there's also still "generate points on a sphere uniformly" ;)

Sp3000

Can we ditch the north pole? :D

feersum

it has probability 0

Transcript for