The Nineteenth Byte

A: Sandbox for Proposed Challenges

Also minified spaghetti code.

New Sandboxed Posts

0

Converting romaji to hiragana A little bit of context first Hiragana is a Japanese syllabary, one component of the Japanese writing system. Hiragana is a kana system. With one or two minor exceptions, each sound in the Japanese language is represented by one character. This may be either a vowe...

12:52 AM

@Downgoat I hate languages that are indent-sensitive.

It looks like the node.js creators haven't discovered else blocks yet..

/shrug

if an if block returns, an else is implied

@Pavel *apt :P

@Zacharý no

Most dpkg-based systems are not what I would consider sane

Better than Windows :P

1:00 AM

Hello? Rasbian user!

@FreezePhoenix Enjoying your Python 3.5?

No. python2.7, and python3.4

Both already installed

Um. You should run apt-get update.

Even Rasbian has Python 3.5 by now

(The joke is that the current version of Python is 3.7)

Um, no, that doesn't work. I'm a Rasbian Jessie user. Jessie only supports node0.10 and python3.4 (I hacked node6.0 on)

Holy shit

How do you survive with such out of date software

1:04 AM

I installed newer versions

> node6.0

I mean, I've got chromium-browser version 54 :P

@Pavel That's what my packages need

jsdoc needs between 6.0 and 8.0

Well, 6.0.0 and 8.0.0

So, naturally, you went with the oldest acceptable version.

/shrugs I don't use node.js - I use vanilla js

That's basically what node is?

1:09 AM

Except for require and module and exports and a whole list of things

@Downgoat Actually there's a reason that code is written like that

@FreezePhoenix You're insane if you use "vanilla" js. "Vanilla" js (AKA pure ECMAScript) has NO I/O/

If you set it up as a not, then you'd need to repeat the ifdef to include the closing brace

Oh wait

No you don't

In fact, they already have it as a regular block...

I'm honestly not convinced that goto was there originally tbh

@Zacharý Ackchyually, it specifies a filesystem api

@Pavel ._._.

1:12 AM

@Zacharý Wrong. Sorry, but I meant browser JS

I just use the browser for I/O tho

@FreezePhoenix Same. Unless I'm hacking something together in JScript on the school computer :p

@Pavel The "Vanilla JS" langauge is maintained and run by TC39. Those FS APIs you're thinking of are either defined by Node or by W3C/WHATWG which browsers implement

I see

Which one of those is ECMA

Grr... 'use strict' makes it impossible to test whether a global object exists

@Pavel ECMA is the specification. Javascript is the language.

uhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh

huh

1:16 AM

TC39 stands for ECMA Technical Committee 39

Ok... anyone got any clues as to how to test if a global exists in strict mode?

to answer your question

@Pavel ECMA-262 === ECMAScript === JavaScript. TC39 === ECMA TC39

~~ill have two number 262s~~

2

At least it's not behind a paywall

1:35 AM

@Zacharý What are you referring to with that statement?

APL

But most APLs are free

And the one that's actually good is

'use strict';
const _global = (0, eval)('this');
const Vector = (function(){
  const isNumber = (a) => { let t = Number(a); return t == t };
  const attemptConversion = (...args) => {
    if( args[0] instanceof Vector ) {
      return args[0].coords;
    };
    if( args[0].every(isNumber) ) {
      return args[0].map((item, _)=> {
        let res = Number(item);
        return typeof res === 'number' && res != res ? 0 : res;
      });
    }
    if( args[0][0] instanceof Array && args[0][0].every(isNumber) ) {

Any clue how to make line #2 less hacky?

Q: Is "Proof-Golf for proofs in general" on-topic?

1:54 AM

@FreezePhoenix do expression pls

New Meta Posts

4:15 AM

0

I recently submitted a challenge proposal on the Sandbox, which is a kind of proof-golf. Here is the background section copied from the proposal: Compass-and-straightedge construction, a.k.a. classical construction, is the construction of lengths, angles, and other geometric figures us...

discussion

4:34 AM

@Pavel I'm talkin' about the ISO specs.

@Pavel yeah.

6:18 AM

Does anyone know of a simple image editing program for Linux
Something that's like Gimp but that isn't Gimp

mınxomaτ

7:10 AM

raster or vector

7:37 AM

What do you all think about embedded development, compared to other kinds of software engineering. Is it interesting? Respected?

code-golf sorting unicode

0

Q: Sort of numbers

Within the recesses of Unicode characters, there exists a Unicode block of (currently) 63 characters named "Number Forms", which consists of characters that have numerical values such as the roman numeral Ⅻ, vulgar fractions like ⅑ or ↉, or weird ones like ↊ (10) or ↈ (100000). Your task is to w...

9:04 AM

@Pavel imagemagik :P

A: Sandbox for Proposed Challenges

9:48 AM

@Pavel Why not GIMP?

New Sandboxed Posts

11:45 AM

0

Einstein's Quicksort code-golf Challenge Given a list of integers and a velocity, \$v\$, implement a quicksort algorithm which simulates a program sorting the list from smallest to biggest at the given velocity. Einstein's Theory of Special Relativity What do I mean? Well assume your program...

FrownyFrog

@Pavel pinta is supposed to be like paint.net but I wasn't impressed, maybe it's got better

try krita

oh, it's for painting, maybe not a good match

Q: Tips for golfing in TeaScript

12:33 PM

0

This question is to list some TeaScript golfing tips

tips

12:59 PM

What's a db or ki:

If there was a DB storing all images containing Blackhat, it would be doable. But do you really expect us to create a KI or so ? — LMD 9 mins ago

Regardless whatever LMD is talking about, I'm not doing anything different than any other image-based test battery challenge

Anonymous

@BetaDecay db = database; dunno about ki

huh I'll ask

Oh sorry - of course in English its AI - Artificial Intelligence... Sorry for using the German Abbrevation... — LMD 2 mins ago

1:31 PM

Unfortunately I don't have time anymore to improve this puzzle, sorry about that. Could someone take it over? — d33tah 2 hours ago

:-/

@BetaDecay There are a very limited number of comics featuring Blackhat compared to your 20, explainxkcd gives 146 comics with him, each probably with multiple panels.

Oh jeez TeaScript. I haven't heard that name in years...

Erik the Outgolfer

the test battery doesn't need to include all of them, does it?

@dzaima Oh

I guess that doesn't really matter anymore

Oh I didn't see that challenge

I should give it a go

1:42 PM

also, I wouldn't suggest having a hidden list of tests as you could be just making up results. Though the workaround is kind of bad too - having the list hidden for a while, posting a hash in the mean time, which does open up hardcoding that while later (and while you could work around that with a competing limit, that's no fun (or maybe an accept time?))

You should definitely set aside some validation data

I see nothing wrong with having a hidden suite of test cases to prevent fitting against the test battery.

I'm going to have a tough time not hard coding (overfitting) if I'm only given 20 pictures and that's what I'm scored on

I'll hide the list until there is a draw and then upload the images I'll use

@quartata You've also got other images on xkcd.com to test with too

@AdmBorkBork if I were the challenge author, I could just say that your answer scored 0, while telling everyone else they've got 100%, which you wouldn't like

1:46 PM

But if you reveal the the test suite after revealing the scores, the score can then be verified

Exactly. It's not a "hidden forever" sort of deal.

@AdmBorkBork oh, I read your message as implying hidden forever as it's after my message suggesting a not hidden forever thing

Oh, I can see how that's confusing.

2:07 PM

y'know, it recently occured to me that, despite using node.js quite regularly, I have never once done any web development with it

I recently had to write a couple hundred lines of JS

But I did not check it into version control.

So I can try to pretend it never happened.

I've become quite enamored with typescript as of late

and the typescript compiler happens to be a node module

I've also been doing some work with discord bots using discord.js recently too

Sounds slow

it's actually pretty fast, I've yet to have a compilation take more than 0.5s

For 1 file?

2:11 PM

for.. 1 sec

42 files

Not bad

granted that's a total of 1726 lines of code

it's fast enough that I can set it up on "watch" mode (recompile every time it detects a source file change) and have no issues at all. Granted Typescript's "compilation" is more transpilation to JS

I was pissed yesterday after spending half an hour to find out that === does not work on strings.

But also, "2" == [[[[2]]]]

So how are you supposed to compare a string anyway?

in js?

Right

2:23 PM

"2" === [[[[2]]]] returns false, what kind of string comparison are you wanting?

But "a" === "a" also returns false.

no it doesn't

at least not in the JS of my chrome browser

Try "a"+"bc" === "abc" then.

also true

same result in TIO's node.js too

Stupid optimizations

I know I had a very simple test case like that that returned false (also in Chrome browser).

And a bug that was fixed by changing my string comparison from === to ==.

2:41 PM

@feersum maybe one side was wrapped - new String("abc")? Otherwise, === on equal strings should return true

4:17 PM

This JS?

4:30 PM

@Zacharý yeah

[] + [] is "" ... JS lacks any common sense.

String("") !== ""

There's no == on arrays for crying out loud!

and then there's JSFuck

A: Sandbox for Proposed Challenges

That's one good thing to come out of JS's "type" system. The other is the wat talk.

To all those interested: RAD's rewrite now supports assignment!

It's good because it can teach about security problems

New Sandboxed Posts

4:47 PM

0

Send the pairs Write two program A and B. A takes 1024 pairs of integers (a,b), where 0≤a<232, 0≤b<1024, and all as are different. Output a positive integer. B take the output of A and one a from the input pairs of A, and output its b. Smallest output of A under a same random test data win.

Adám

@Zacharý RAD on TIO when?

@Adám After I get a few more things done: namely representation (we don't want 0 to be 0+0i), a few more essential atoms (I still don't have a lot of structural atoms), and reading code from a file.

Never mind about assignment being fully supported: still have to fix stuff like A←1 2 3⋄A[1], which currently breaks the interpreter

Neil

5:13 PM

phew, for a moment there I thought I wasn't going to have the shortest Charcoal answer on Beta Decay's ascii-art question

@Zacharý well it kind of makes sense - it's pretty useful that "2+2 = " + (2+2) works, casting the args to strings. The string representation of an empty array is "", so ""+"" = ""

@Zacharý String(...) is calling the String class constructor function without the new keyword so you're gonna get strange stuff, besides, what did that do that you didn't expect?

@Zacharý arrays are objects. If I did arr.foo = "bar" on an empty array, should [] == arr?

5:33 PM

I know the reasoning behind it, it's just weird.

I don't think I understand this language

Have you all seen images.law.com/contrib/content/uploads/documents/394/2560/…

IBM suing Groupon for stupid shit

I still need to fix indexed assignment, but I got A←1 2 3⋄A[1] to work

5:55 PM

@Poke All I can tell is a patent is being infringed

@Pavel It boils down to IBM suing Groupon for using SAML with user provisioning and selling things online

there are like 3 different patents

I'm not really seeing why it's stupid tbh

SAML existed for 7 years before IBM patented "user creation during SAML"

like just tack a thing on and file a new patent

Ah

So the patent itself is dumb

so my gripe is not so much that the lawsuit is ill-found

yes

that the patents are super dumb

The other one is like... creating session state over a stateless protocol (http)

which is what HTTP Cookies are

The actual counts on on pages 9-15

6:04 PM

I wonder if I can use Fedora Media Writer to burn an OpenSUSE iso

6:49 PM

I have no clue how to do indexed assignment in RAD.

7:07 PM

@Zacharý you can desugar it as amend

Amend is @, right?

@Zacharý yes

@Zacharý You mean like foo[x] = whatever?

@Pavel Yeah.

@Zacharý If every object has a pointer to itself, the operator = could be handled by the object itself modifying the data at its own address.

7:13 PM

@ngn I'm concerned about how ; would behave there.

@Zacharý the separator between indices? as in a[i;j]←b?

But that will work once @ is implemented, even if it is slow as heck.

@ngn Yes, ; is an index separator. I'm not that concerned about A[I;J]←B with I and J being scalars. I'm mostly concerned about how something like A[1 2;2 1]←B would get desugared.

@Pavel I had the same issue with trying to overload opIndexAssign. I had to basically do what @ngn said and do ammend on it (albeit, it was in D):

	public Value opIndexAssign(T, U)(T value, U index) { //TODO: SPEED THIS UP
		if(this.peek!(Value_[]) is null) {
			throw new Exception("RAD: INTERNAL ERROR");
		}
		Value_[] res;
		for(int i = 0; i < this.length; i++) {
			if(i == index) {
				res ~= Value(value).theValue;
			} else {
				res ~= this[i];
			}
		}
		if(index == this.length) {
			res ~= Value(value).theValue;
		}
		this = Value(res);
		return Value(value);
	}

@Zacharý that depends on how you define your amend, A[1 2;2 1]←B could be a combination of amend-each-amend or it could use a separate primitive for dmend (amend at depth)

but I wouldn't worry about it on first implementation :) a[i]←b is complex enough

Okay, I'm going to have to go through the same hell as I did with ;. I tried it in so many different ways, none worked correctly: I had to port the Python version and reduce using a constructor.

@Pavel AFAIK, RAD has no refcounting. I don't think he can modify in place, or else a←b←1 2 ⋄ a[1]←3 ⋄ b would be wrong.

7:24 PM

I'm pretty sure that's how C++ implements =

But IDK

@Pavel C++ has reference semantics, sane languages have value semantics - in the above example b should remain unchanged

7:37 PM

Part of my problem is just entirely my fault: ← doesn't even handle indexes right. It just assigns the name right away, so the namespace will have the inaccessible A[1] when A←1 2 3⋄A[1]←2 is called.

Neil

7:48 PM

that feeling when you golf over half the bytes of your answer because it was so bad before

8:03 PM

@Zacharý I've no idea what your last message means, but I think I should follow your lead and implement a[i]:b in ngn/k :)

Huh... github.com/GaloisInc/cryptol/issues/470

@ngn And that'd be wrong regardless, anything that starts with lowercase is a function.

@Zacharý my examples are in APL and k, I'm not competent enough to write proper RAD

@ngn That's one of the only real differences, that along with ⍤ being depth rather than rank, and / being somewhat switched with ⌿ (scans too).

; indexes by depths as well. RAD isn't really different from APL (symbols) and K (rankless).

8:29 PM

@Zacharý well, if case matters for something being a function, it is different from both...

8:41 PM

I think I found out how to implement indexed assignment without using @.

@Adám, this gives me an idea for a new definition of ε⍵: εnumerate.

ε(1 2 (3 4 5)) => (,1) (,2) ((3 1)(3 2)(3 3))

Cryptol is a really interesting language. In many languages, a string is just an array of chars. In Cryptol, an int is just an array of bools, so 12 # [False] == 24 (where # is the concatenate operator)

@Pavel Cryptol has something to do with designing cryptography algorithms, right?

8:57 PM

@Potato44 Yeah, it's entirely a cryptography language. It has a lot of emphasis on manipulating bits, and it also has support for theorem solving.

It outputs everything in base 16 by default, and doesn't support negative numbers or floats.

I think I had a look at its website for a few minutes when I was looking for languages for the polyglot.

It got onto TIO yesterday

I don't think it'd be good for polyglotting

no, that's why it was only a few minutes, but it seems like an interesting language otherwise.

and one with a specific purpose

I don't think it can even do IO tbh

It just implicitly prints the result of every expression

now I'm imagining Sha1, but a version that outputs all the intermediate states over http

9:01 PM

Cryptol> 1+1
0
Cryptol> 1+1:[8]
2

When you don't specify a type, it assumes 1 is a 1 bit integer so 1+1 overflows. 1:[8] specifies 1 is an 8 bit integer

is + xor?

+ is integer addition

ah, size specifier that default to 1 bit

So 2+2 is also 0, but 3+3 is 2.

Since 3 is a 2 bit integer by default.

Do you mean 2-bit by default? otherwise that doesn't make sense

JungHwan Min

9:04 PM

VTC? codegolf.stackexchange.com/q/169164/60043

Yes

2 bits

JungHwan Min

Dang, that was fast

Q: Generate an expression yielding near-equidistributed result

9:17 PM

1

It is asked a program (any language) which, given an integer input m in range [1, 65535] outputs one or more C expressions that all: Uses a single variable x assumed of 32-bit unsigned type (e.g. uint32_t or unsigned ) Would evaluate to range [0, m) for any of the 232 possible x, reaching any o...

code-golf code-generation

It probably assumes the lowest number of bits possible to represent the integer.

Right

Question you probably can't answer: is 0 zero or one bit?

It won't make a difference, since it's zero.

@Zacharý 0 is zero-bit. Harder question: how many bits is -1?

In cryptol?

9:20 PM

-1 is type {a} (fin a, a >= 1) => [a]

._.

@Pavel er... wat?

[a] means a-bit int, and a is specified to be finite and >= 1.

So -1 is 1 bit.

The amount of space values take up must be insanely small.

0 is {a} (fin a) => [a], so it is in fact 0 bits

The type of "foo" is [3][8], a length-3 sequence of 8-bit ints.

So "foo" and "foobar" have different types

{k} (2 >= k, k >= 5) => [k] is a valid type. That said, not value can actually have this type, but still.

9:38 PM

Speaking of types: imagine a language where types are first-class objects. so stuff like typeof(member)[number] value; is possible. That'd be one weird language.

I was just thinking that, but types are most useful for compile-time checks

That wouldn't be possible with first-class types

@Zacharý I'm having a hard time deciphering that syntax, but that sounds vaguely like dependent types. At least the first class bit

10:12 PM

It could be interpreted as well, making it even weirder.

It might be possible for type errors to occur at both run-time and compile times, depending on whether the type can be determined at compile-time or not.

@Zacharý is this the kind of thing you are looking for?
https://tio.run/##lVHBSuRAEL33V7zNxQQ24kEvgRECLqwgehhxDyLSJJWZhqQ7dnfGEZmbv@AP7o/MVqcjRj3IhtBd3e9VvXrVclXL/d70pKG63liP8ysxP55JLw8vpRdC5DkulPMOpkEvrVfV0EqLlvTKr5FuqPLGukzUnIIbqpCWKLAkn/H29@UV@Wk44XFNlgRwewcwEpgljvjiviju@eJZR/4uJJRhiRT9HqZuqKAzIZRu1NbiZMwdW/y19VZW3KV/6gmNNR02sh0IUteoLElPjK0JgVwZa8n1RtdKrxAdxET26HqqVKOoniwKxnPT5CPObU7udumWgzJLp7azyecH@nO5w5YdLN68xHkuvZ0Js@Y0zGOkIS0TbiTkodQkGybI3xdkgblecr1WDvxLRGLCNb@THOc012S5TzJzcIGkTFAUSGKxGJsm7pHoxsPtXXyah0GxBmkPb4pZqQMW@r77Of//tIP42dB1T@ik0iIsXMEOGmk/eJ4Ikt/UtuYn/hjb1j@SbL//Bw

10:29 PM

Something like that. It looks like Haskell

H.PWiz

It says "HASKELL" in the debug

Yes, Agda's syntax is inspired by Haskell

(For what it's worth, Cryptol is written in Haskell)

@H.PWiz yeah, that's because some part of the standard library is apparantly using a depreciated part of the Haskell FFI

I have no clue what I'm doing wrong: on my local copy of RAD, A[1]←1 does A[1]←⍬ for some reason

code-golf number decision-problem rational-numbers

10:38 PM

0

Q: Mixed Fraction Equality

In elementary school, children learn about proper fractions, where the numerator is less than the denominator, and thus the value of the fraction is less than one. Later, they are taught about fractions where the value of the fraction is greater than one, and two different ways to express these f...

Version with syntax closer to what Zacharý used as example.
https://tio.run/##lVFNi9RAEL33r3jmYgJGPOglMEJgFxREDysKDsPQJJVMQ6Y7dnfGGZa5@Rc871V/hCf/yfyR2ep01OgeXEPor3pV770q2dbyfDY9aahtb6zHyzdifr2QXj5@Lb0QIs/xSjnvYBr00npVDZ206Ei3foN0R5U31mWi5hS8owppiQJX5DPeTp@/IH8ebvi0IUsCWK4AjgRkiSf8sC6KNT9c64g/hoQyLBGifx9TN1TQmRBKN2pv8WzMHSVe7r2VFav0h57QWLPFTnYDQeoalSXpiWMbQgBXxlpyvdG10i2ig5jIHl1PlWoU1ZNFwfHcNPkYZ5mTu2O650OZpZPsbPL5B/y6PGLPDhY/vcR2Euts1Y6gPNKqM47gTTbKcwft5R4fZLWR9sd3uKFtyXmqRShomtPN1/Xp5ttyvbqPljs5C8zkjWKuvJ01gf1Pg32KNGAy4UZAHvKmFoRp8ncnssAvPiRvN8qBf4kISxAkcNnVv2jHuc15mfIvqnlwgaRMUBRIYrF4Nk3…

11:37 PM

@quartata have you tried mixing emscripten and v8 before