The Nineteenth Byte

PhiNotPi

00:00

I've seen some really bad kerning before.

hosch250

So have I, but scrubbing burned oil is worse.

Peter Taylor

The pain in your wrist only lasted for days. The ability to recognise bad kerning lasts for a lifetime.

3

hosch250

It sure hurt more than a lifetime of bad kerning.

Sorry to go, but supper is going on.

PhiNotPi

"When you gaze long into bad kerning, the kerning also gazes into you"

hosch250

What is that quote from?

Martin Büttner

00:04

@PeterTaylor You know JS, right? That 10-byte one is puzzling me.

PhiNotPi

It supposedly originated from Friedrich Nietzsche.

Martin Büttner

I just noticed, in the long one, I can make flush from the remaining characters... but that doesn't seem to be a thing in JS

PhiNotPi

Which program are you trying to unscramble?

Martin Büttner

all three by user Cris

codegolf.stackexchange.com/a/41099/8478, codegolf.stackexchange.com/a/41101/8478, codegolf.stackexchange.com/a/41104/8478

PhiNotPi

00:27

I'm surprised how active the challenge is.

It seems to be a lot essier for the robbers than anticipated.

grc

@MartinBüttner the 10-byte one looks to me like it relies heavily on its environment (which we don't know)

Martin Büttner

@PhiNotPi there haven't really been any submissions relying on factorising large semiprimes

@grc probably... I'm currently trying that PHP one

by bwoebi

49 bytes

I'm pretty close actually

PhiNotPi

10! = 3628800. I think bruteforcing could be possible. Like you said, however, we don't know the environment.

xnor

my 12-char python one was brute-forced, though there were many equivalent solutions

Martin Büttner

for the PHP one I've got print@substr(new exception,); and I need to put some permutation of $++$++$++$+=$~main$2 after the comma to evaluate to -6

Martin Büttner

00:45

ohhh

or maybe not...

hm that's odd

I thought about doing something with variable variables

since all of $m, $a, $i, $n are NULL, the variables $$m and $$a are the same.

(and I've tested that)

but $$a+$$m=2 yields 2 and not 4

as does $$m+$$m=2 ... whereas $m+$m=2 does yield 4...

PhiNotPi

01:10

Did you know that $$m+$$m=2 gives an error in Java?

~

Martin Büttner

lol

PhiNotPi

It occurs nearly 1000‰ of the time.

grc

01:23

print@substr($m=new exception,); gives you a variable which becomes 1 if it helps

Martin Büttner

@hosch250 how is that python answer coming along? ;)

hosch250

Almost done.

Martin Büttner

@grc hmmm, let me think

hosch250

Actually, just finishing killing the final unwanted features.

Done!

Now to golf it.

Martin Büttner

@grc the closest I get with this is print@substr($$m=new exception,~(++$$a+$$i+$$n+2)), which has two parentheses too many and two +s missing. (but does give the correct output)

okay, I can just add the +s before 2

but how do I get rid of the parentheses?

actually, it also has two extraneous $

chilemagic

01:34

hey guys working on unscrambling this answer:

0

A: Unscramble the Source Code

Perl, 39 Code //shredding($c_)=\\t_;$_=($r/~$p*2014); Output H Tested here

$$2=__g
$_=0;
$_= ~$_;
s/(\d\d)/print(chr$1*4)/e;

im sorta close

feel free to steal my work and get a head start, I don't think I'll figure it out

Martin Büttner

I don't really know perl

chilemagic

Martin no worries maybe @grc can get it!

also to throw out my thoughts about this one:

0

A: Unscramble the Source Code

Perl 5, size 27 print(rar(pizza))=~--'tanz' Output: ohm Works under use strict.

I think he might be using tr/// with z as a delimiter

grc

@chilemagic how did you end up going with this one?

hosch250

Up, one more bug.

chilemagic

@grc I had one other idea about it being a for(;;) loop

hosch250

01:39

bugCount+=1

chilemagic

where you set $_=r or something and then do ++$_ to cycle through it

think of this type of thing:

for($_=r;$o<500;++$_){
print if /pe/ or /rl/;
$o+=1;
}

so yeah that or for(fi..ro)

could get you pe and rl

Martin Büttner

okay, I've almost got it now

hosch250

bugsRemoved+=1

Martin Büttner

@grc @PhiNotPi

0

A: Unscramble the Source Code [robber thread for cracking attempts]

PHP, size 49, by bwoebi print@substr(new exception,+~$$a+=++$$m+$$i+2+n); That was absolutely mental.

grc

nice!

Martin Büttner

02:01

yay, another Mma answer! :)

Sp3000

Man, codegolf.stackexchange.com/a/41098/21487 makes me feel like I'm solving a Kolmogorov

bwoebi

oh, you were looking at my scrambled source here in chat :-D

hosch250

Posted, @MartinBüttner.

Martin Büttner

first thing for golfing python: first indent is one space, second indent is a tab

40

A: Tips for golfing in Python

Sometimes your Python code requires you to have 2 levels of indentation. The obvious thing to do is use one and two spaces for each indentation level. However, Python considers the tab and space characters to be different indenting levels. This means the first indentation level can be one space...

bwoebi

@MartinBüttner anyway, well done :-)

Martin Büttner

02:16

@bwoebi thanks :)

Sp3000

Py 2 though :/

grc

@chilemagic I've cracked the Perl 39 one. Do you want to post it since you were so close?

Sp3000

print "\n".join(chr( if (x+2)%3 else )+"#4"[(x+2)%3] for x in range(7))

This is hard with no 0 or 9 :/

hosch250

@MartinBüttner Are you sure? I just did this, and my two programs gave different solutions.

Is it because I have more than one level of indent?

Oh, I am using Python 3 - see the comment.

That is the problem.

grc

@hosch250 a single tab is considered 8 spaces in Python 2/3

hosch250

02:25

@grc OK, either way, tabs are less than spaces byte-wise and char-wise.

grc

it means you could have indentation levels 1 space, 2 spaces, 3 spaces, etc. or 1 space, 1 tab, 2 tabs, 3 tabs, etc.

but you can't go back to spaces after a tab

if that makes any sense at all

hosch250

I did only use space for the first level of indentation, it didn't work.

Also, did you see this comment on the post?

This fails in python3: you can no more mix spaces and tabs(a bad thing for codegolf, but a good thing in all other cases).

It has 18 upvotes, so it must be correct.

Martin Büttner

@grc couldn't you do s, t, st, tt?

grc

@MartinBüttner I think t == st

(I did an assignment trying to parse Python indentation)

(it was painful)

Martin Büttner

and ts?

grc

02:29

ts -> 9 spaces so it would work

See this

Martin Büttner

yeah that's what I meant, I guess

anyway, I'm off to bed... leave me something to crack for tomorrow, but feel free to get rid of all that are shorter than 20 bytes :P

grc

gn

Sp3000

ideone.com/h80URb Here's Py3 at least

ideone.com/jcMkdA And Py2 seems to work

grc

@Sp3000 guess it would only work in Python 2 then

hosch250

Well, I will try it again.

bwoebi

02:32

Now I need to invent a new one for @MartinBüttner to crack… hrmpf! (PHP is btw. a very nice language with its tiny quirks :-D)

grc

tbh I usually just use spaces because tabs are annoying

even when it costs a few chars

Martin Büttner

@bwoebi ugh, please don't :D

Sp3000

I think my IDLE autoconverts tabs to spaces

bwoebi

@MartinBüttner not before tomorrow… I go to bed now too. Good night

Martin Büttner

I spent way too much time on this, and I should focus on the SAS and APL ones, because those can beat me ^^

Sp3000

02:33

Golfscript :P

Martin Büttner

oh and that

Sp3000

GL with that one :P

Martin Büttner

lol

I'm sure you'll have it before I wake up ;)

good night

Sp3000

(I don't think so but night :P)

bwoebi

@MartinBüttner Such languages usually look less like gibberish to me when scrambled than unscrambled

hosch250

02:37

OK, everyone. I just replaced all my tabs with a single space, and it is the same bytewise.

Sp3000

:P

hosch250

No luck there, and I won't bother editing my post.

Sp3000

Is this for rivers?

hosch250

Yup.

grc

@hosch250 if you look into list comprehensions, you could save a lot of characters

Sp3000

02:38

Rather than "a is b" you can do "a==b" to lose the two spaces

hosch250

OK, thanks.

Need to go do chores soon, but will remember this.

Sp3000

I wonder if you can do something like sum(j<len(l[k])and' '==l[k][j]and j<len(l[k+1])and' '==l[k+1][j]) for the big nested ifs chunk

Cos that's a lot of indents

grc

02:57

or even o=[sum(j<len(l[k])and' '==l[k][j]and j<len(l[k+1])and' '==l[k+1][j]for k in range(len(l)-1))for j in range(i)]

hosch250

@Sp3000 Just found that on my own :)

Sp3000

Also q+=[max(o)]?

I think you're finding the max, right?

@hosch250 Ahaha nice

hosch250

Yup, just solve that too: q.append(max(o))

Would q+= be the same thing?

Sp3000

Yeah :P just adds two lists and is usually shorter than append

hosch250

OK.

Thanks.

Sp3000

02:59

i.e. you can do a+b if a,b are lists rather than a.extend(b)

hosch250

I had to use .append.

The compiler wouldn't accept +=

Sp3000

Did you put [] around the variable?

hosch250

Oh, OK.

Sp3000

Yeah, need to turn it into a list :P

hosch250

OK, worked.

What about .split?

Can that be shortened?

grc

03:02

t=input().split()

unless you need input() again

hosch250

OK, thanks.

Sp3000

But otherwise in general I don't think .split can be shortened, not that I know of

hosch250

437 bytes.

Sp3000

I'm look at the last line - is there anything shorter than print(*l,sep="\n")? It's only one char shorter

chilemagic

@grc go ahead and post it! mention that I helped though :)

hosch250

03:06

What about getting the length of a string?

Rainbolt

.len()

hosch250

That would be longer.

I am using len()

Sp3000

You can redefine x=len and then reuse x() if you use len a lot

Rainbolt

OH it isn't a member function. Ignore me

hosch250

Yeah, I use len a lot.

Redefined range to x too

417 bytes

Sp3000

03:12

That's pretty good so far :P

hosch250

Is there a function for index of?

I could use that.

Sp3000

.index()?

That gives the first occurence

If you want the last I think there's .rindex()

hosch250

Either is fine.

I had been using the last to avoid a break.

30 bytes :)

Sp3000

:D

hosch250

Shaved off more: print(*p(i+70),sep='\n')

FryAmTheEggman

03:16

find() is usually shorted than index()

Sp3000

Nice :P

hosch250

List has no attribute find.

FryAmTheEggman

Ah, I should have looked further back, use find() on strings

my bad :S

Sp3000

I keep forgetting find exists :P

hosch250

Can I shorten this? for i in b(70,91):

I have a few of these still.

for i in b(i):

Sp3000

03:18

Do you want to post your code so far so we can see where you're up too?

hosch250

Sure.

Sp3000

Might be able to list comprehension as grc said, depending

hosch250

384!! :)

Sp3000

:D yay

(I'm always impressed by Python golfs, just because it usually never wins and I like Python)

hosch250

Let get it to 99 :)

Or lower.

Sp3000

03:20

XD

hosch250

Still the longest solution, but not by nearly as much.

Does Python have the & or && operator for and?

Sp3000

I like the def p(i)

Hmm & is bitwise

hosch250

I did that because I used that twice.

Sp3000

What you can do though is swap l[k][j]is' ' around

To ' '==l[k][j]

So that you can lose the spaces on either side

hosch250

I don't see any spaces.

if j<x(l[k])and j<x(l[k+1])and l[k][j]is' 'and l[k+1][j]is' ':r+=1

Sp3000

03:23

Oh on the left side, sorry

and l[k][j]is' 'and -> and' '==l[k][j]and

But yeah as grc pointed out you can list comp that into o=[sum(... for k in b(x(l)-1))for j in b(i)] where the ... is the long if condition

That covers everything from the line o=[] to o+=[r]

FryAmTheEggman

j<x(l[k])and j<x(l[k+1]) could be x(l[k+1])>j<x(l[k])

hosch250

Thanks, @FryAmTheEggman

Sp3000

Ahaha comparison chaining, gotta love Python for that :P

hosch250

I do.

FryAmTheEggman

:)

Also could all 3 of those ever be false?

You might be able to use == instead of and

Sp3000

03:28

You have so many you could almost use all(), but I think that'd end up longer after the comparison chains

hosch250

Ah, l[k][j]is' 'and' '==l[k+1][j] becomes l[k][j]is' '==l[k+1][j]

FryAmTheEggman

Yeah just be careful with ==, you have to be sure that both won't be false at the same time

But otherwise its great! It even short circuits

Sp3000

Why would them both being false at the same time be a problem?

FryAmTheEggman

False==False gives true

Whereas False and false is false

Sp3000

>>> a = 1
>>> b = 2
>>> c = 3
>>> a == b == c
False

Hm...

FryAmTheEggman

03:31

Yeah :)

Sp3000

I don't get it :/

I thought chaining just made (a == b == c) -> (a == b and b == c)

FryAmTheEggman

If you have bool1 and bool2, you could write it as bool1==bool2

Sp3000

Which is true iff all three are equal?

FryAmTheEggman

But only if bool1 and bool2 are never both false

Try 0==1==asdfafsafasfa

hosch250

This is what I have: l[k][j]is' '==l[k+1][j]

Maybe the is will prevent that?

Sp3000

03:32

>>> 0==1==asdfafsafasfa
False

That looks okay to me

FryAmTheEggman

Yes, that's what it should do

hosch250

339 bytes.

FryAmTheEggman

But False==False==False is True

Sp3000

Well isn't that what you'd expect though? They're all the same

We're checking if three things are the same, right?

FryAmTheEggman

Yeah, but False and False and False is False

Sp3000

03:34

(also about is on strings: stackoverflow.com/a/2987975/2771429)

hosch250

That would be like 0==0==0.

FryAmTheEggman

Basically the golf I suggested is not 100% safe

hosch250

But we aren't really using and except to chain statements, I had essentially x==0 and y==0.

FryAmTheEggman

I think I'm bad at explaining this :S

Sp3000

I'm just a little confused at how it applies here, sorry :(

hosch250

03:36

So x==0==y should be the same thing.

Sp3000

^^ that was what I was thinking

hosch250

It isn't quite like a and b and c.

FryAmTheEggman

OK I just wanted to make sure you thought it through

I have no idea what it is actually doing

:D

Sp3000

Ahaha k XD

grc

(False==False==False is True) is False :P

Sp3000

03:37

btw @hosch250 you can move i from the second last line into the last line if you haven't done so already

hosch250

338 bytes.

I have i=q.index(min(q));print(*p(i+70),sep='\n')

Sp3000

I mean print(*p(q.index(min(q))+70),sep='\n')

FryAmTheEggman

>>> False==False
True

?

Oh

hosch250

3 more chars: print(*p(q.index(min(q))+70),sep='\n')

grc

>>> False==False==False is True
False

hosch250

03:38

@FryAmTheEggman You are doing equality, not and

Sp3000

We're comparing if things are the same :P i.e we're not doing (a==b)==(c==d), just a==b==c==d

hosch250

>>> False==False==False
True

FryAmTheEggman

>>> (False==False==False)is True
True

I think precedence might screw it up?

hosch250

Yeah, probably.

grc

the 'is' operator chains

I'm just being annoying, ignore me

FryAmTheEggman

03:39

Ah :P

Sp3000

You know, I wonder if the middle chunk can all go into a massive nested ugly list comp

FryAmTheEggman

Yes that is correct

hosch250

334 bytes.

Posting

FryAmTheEggman

Forgot to update byte count ;p

hosch250

Done.

OK.

Still the longest :(

99 bytes to go to tie up.

Sp3000

03:44

q=[max(sum(x(p(i)[k+1])>j<x(p(i)[k])and p(i)[k][j]is' '==p(i)[k+1][j]for k in b(x(p(i))-1))for j in b(i))for i in b(70,91)] saves a few chars but runtime goes massively downhill so I'm not sure if that's a good idea :P

FryAmTheEggman

The final for loop can be one-linered with semicolons

Sp3000

... that's probably a better idea :D

FryAmTheEggman

:P

grc

for i in b(70,91):l=p(i);q+=[max(sum(x(l[k+1])>j<x(l[k])and l[k][j]is' '==l[k+1][j]for k in b(x(l)-1))for j in b(i))]

this should work too: l+=[z]*(z!=l[x(l)-1]);return l

Sp3000

Woo almost 300 and getting unreadable, now this feels like golfing :P

grc

03:46

or return l+[z]*(z!=l[x(l)-1])

FryAmTheEggman

you could define l and z as named arguments in p

p(i,z='',l=[])

Sp3000

Errr be careful with l

grc

I think that would only work for z

Sp3000

Mutables

FryAmTheEggman

right

hosch250

03:49

I don't even know where you guys are anymore :(

grc

last two lines of function p can be changed to: return l+[z]*(z!=l[x(l)-1])

FryAmTheEggman

... And that's why you never use what you learn here for production code :D

hosch250

Yeah.

grc

and you don't need the semicolon at the end of the first line

hosch250

Yeah, I had deleted something else.

I think.

Gotcha now.

315 chars.

:)

Sp3000

03:51

Gogogo beat bash :P

hosch250

Fat chance, I think.

I'd like to.

Is there anyway I can combine these lines?

t=input().split();q=[]
for i in b(70,91):l=p(i);q+=[max(sum(x(l[k+1])>j<x(l[k])and l[k][j]is' '==l[k+1][j]for k in b(x(l)-1))for j in b(i))]
print(*p(q.index(min(q))+70),sep='\n')

grc

the slower method mentioned before by @Sp3000

hosch250

OK. Hold on.

Sp3000

This is probably terrible, but maybe l,z=[[l,z+n+' '],[l+[z],n+' ']][x(z)+x(n)<=i] after for n in t: ?

xnor

@hosch250 I noticed your answer contains if x(z)+x(n)<=i:z+=n+' '

hosch250

03:55

OK.

xnor

in general if b:s+=t can be replaced by s+=t*b

hosch250

Hold on, I just cut 2 chars from p:

 for n in t:
  z+=n+' 'if x(z)+x(n)<=i else l+=[z];z=n+' '

I'm losing track again.

xnor

erm

i don't believe you can do assignments in a ternary

Sp3000

I think that does something different

hosch250

311 bytes.

@Sp3000 No, I've done this a lot before.

a if cond else b

grc

03:57

@Sp3000 1 char shorter, assuming I got the order right: l,z=[l,l+[z],z+n+' ',n+' '][x(z)+x(n)<=i::2]

hosch250

Instead of: if cond: a else: b

Sp3000

No I mean, does this get parsed as z+=n+' 'if x(z)+x(n)<=i else (l+=[z];z=n+' ') group the last or (z+=n+' 'if x(z)+x(n)<=i else l+=[z];)z=n+' '

But also assignments in ternary yeah

grc

I don't think Python allows assignments in expressions at all

hosch250

OK, it crashed.

That hadn't worked.

FryAmTheEggman

RIP program, you will be remembered.

hosch250

03:59

I've done that before, but it must have been somewhat different.

I have it working again.

FryAmTheEggman

(Is thrown to the garbage collector)

hosch250

315 bytes again :(

Sp3000

>>> x = 1
>>> x=' ' if x==2 else x+=3
SyntaxError: invalid syntax

Yeah I don't think you can do assignments

In a case like that you can do x=' ' if x == 2 else x+3, which doesn't help our case

hosch250

Is there any way to shorten z=n?

I don't think so.

grc

does this work: c=x(z)+x(n)<=i;z=z*c+n+' ';l+=[z]*(1-c)

hosch250

04:03

Where?

Sp3000

Oh my

grc

for n in t:

Sp3000

grc's pulling out the insanity

hosch250

Ooh.

grc

we have to beat bash right?

hosch250

04:04

name n is not defined.

grc

for n in t:c=...

hosch250

Nope, I don't know what it does/

eismod
minim
aliquip
reprehenderit
pariatur.
culpa
dolor
incididunt
quis
commodo
velit
occaecat
mollit
mollit anim id est laborum.

This is the output.

314 bytes :)

Oh boy, I've wasted over an hour here!

I HAVE to go NOW!

grc

ikr... I have an exam tomorrow :(

Sp3000

I'm meant to do an assignment but... nah

hosch250

I have chores to do - it is past 10PM!

Sp3000

04:07

:P thanks for the fun

hosch250

You too!

@Sp3000 I just tried your de-optimizer, and it didn't save a byte.

Sp3000

Which one?

hosch250

So we don't need to worry about that.

q=[max(sum(x(p(i)[k+1])>j<x(p(i)[k])and p(i)[k][j]is' '==p(i)[k+1][j]for k in b(x(p(i))-1))for j in b(i))for i in b(70,91)]

Sp3000

Ahaha k

Yeah I think once we moved only max in it didn't help much

hosch250

:18497543

Sp3000

04:10

(because l got expanded to p(i))

hosch250

OK. Thanks for helping!

Bye!

Sp3000

:) k

Well damn Martin's got cracked

Transcript for