The Nineteenth Byte

and we have shifted to a SBCS, but we haven't created a single language that uses the taboo of non-byte-based encodings (huffman/decimal (decimal?) would be the best)

Adám

@ASCII-only Hm, I guess it wouldn't be so hard to target ten specific challenges with a single language winner (e.g. Jelly) and just add enough domain-specific single-byte built-ins to exactly best those ten winners by 2 bytes. It just requires combining to built-ins two a single byte twice per challenge, meaning adding 20 built-ins. I'm sure one can find 20 built-ins that are not needed for those 10 challenges.

ASCII-only

well. IMO just using 05AB1E + more builtins + decimal encoding alone can fit the task, for 10 if not like 40 of the challenges

lirtosiast

you all raise some good points

Adám

@ASCII-only However, it may of course count as sufficiently cheaty to be disqualified.

ASCII-only

12:13 AM

cheaty?

lirtosiast

I'll disqualify a language if it adds builtins that aren't useful elsewhere to win the 10 challenges

Adám

@ASCII-only One of the criteria is that the language isn't cheaty. We're basically devising a strategy here to cheat on the bounty spec.

lirtosiast

like

Adám

@lirtosiast It will be hard to claim that a combination of just two built-ins into a single built-in isn't useful elsewhere.

lirtosiast

my goal with this bounty was to incentivize creating a language that would win 10% of all challenges by a significant margin

ASCII-only

12:15 AM

IMO that's not good enough

lirtosiast

And I'm not going to make anyone write more than 10 programs

Adám

I wonder what the most common two-character sequence is in Jelly.

lirtosiast

So we're stuck with a limited sample size

ASCII-only

20% of challenges by 20%, 50% by 10%, or 10% by 40% maybe

@lirtosiast keep in mind that there are very, very, very many simple challenges

which the language should be able to do in under 20 bytes

lirtosiast

True...

Adám

12:17 AM

Just taking Jelly and replacing the N least used single-byte built-ins with the N most common two-built-in phrases would probably qualify.

ASCII-only

Yeah. I'd increase the threshold until there are some attempts, after which you can revise your percentages

We don't know how easy/hard it is to beat a certain percentage of challenges bby a certain amount

lirtosiast

@Adám Jelly has already had aliases added for 2 byte builtins so the gains would be limited

Adám

@lirtosiast No I mean single-byte aliases each for two 1-byte built-ins combined into a single built-in.

lirtosiast

@ASCII-only OK I'll change to 20% by 20%

ASCII-only

@Adám Jelly's codepage is full already though?

Adám

12:20 AM

@ASCII-only Plenty of less-used built-ins could be removed to win this bounty.

ASCII-only

yeah

or even just switching the builtins around to make Jelly the best for a different subset of challenges

while still retaining its full capabilities

Adám

Yup.

lirtosiast

@ASCII-only Jelly's codepage isn't quite full. There are 10 or so characters still unassigned.

ASCII-only

O_o

lirtosiast

Which gives you another 1% decrease in code size

Veskah

12:23 AM

There's always room for the Hello World built-in

ASCII-only

:|

that is the most useless builtin ever

ngn

12:58 AM

github down?

Pavel

@ngn Up over here

Dennis

@ASCII-only Try it online!

2

ngn

up again for me too. guess i caught it in a bad moment

ASCII-only

3:31 AM

@Dennis Try it online!

Dennis

And how exactly is that more useful than Hello World? :P

ASCII-only

random question: what should happen if you assign to object.foo if it's a property of the class (and not the object)? what about class.foo if it's a property of a superclass (and not the class)?

Dennis

I'd expect an error.

ASCII-only

@Dennis in my (our?) defense, they're not builtins, rather predefined variables. plus, they're not a single byte long

Downgoat

@ASCII-only property of class meaning it is static field?

ASCII-only

3:35 AM

@Downgoat well kinda

Dennis

If you want to make them mutable, you have to copy all class properties for every instance or add a whole bunch of conditionals.

ASCII-only

assuming functions are first class objects, it's hard to tell methods from properties

@Dennis solution: copy on write/fallbacks (fallbacks = conditionals)

Downgoat

@ASCII-only 1) object.foo should not accessible unless doing class.foo or typeof(object).foo 2) class.foo should modify property of superclass

ASCII-only

oh yeah first another question: if functions are first class, should methods be treated basically the same as properties?

Downgoat

@ASCII-only in terms of assignment semantics yeah

though if you are doing that I'd recommend a readonly attribute or something of the sort

Dennis

3:39 AM

@ASCII-only I'd still go with an error. If you're changing class properties, you're making a new class.

ASCII-only

so... now assume object.foo is a property/method (especially method). what now

@Dennis what if it's a scripting language

oh wait

Downgoat

though JS object semantics are a mess to be fair...

ASCII-only

true. you either modify the class itself, or create a new class

Dennis

It's still a new class. Having the same properties is what defines it.

Downgoat

also quick question: how are interfaces aligned in memory. Do interface v-tables have a pointer to the beginning of the struct?

optimizing fixed interface v-table offset alignments for all interfaces globally seems like a very complex task

Dennis

3:41 AM

I guess dynamically creating a new class would be an option, if changing class properties has to be possible. I doubt that's a great idea though.

Changing the class's own properties via an instance is most likely an accident and shouldn't be allowed.

New Main Posts

4:07 AM

0

Q: Increment a GUID

Inspired by a recent Daily WTF article... Write a program or function that takes a GUID (string in the format XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX, where each X represents a hexadecimal digit), and outputs the GUID incremented by one. Examples >>> increment_guid('7f128bd4-b0ba-4597-8f35-3a2f27...

code-golf

Anonymous

@Downgoat Huge understatement

Anonymous

@cairdcoinheringaahing related

ASCII-only

4:24 AM

@Dennis if you're talking about class modification/reopening (ruby-style): what if you want to change the behavior of primitive/builtin types?

@Mego :| no it's a bit of an overstatement

probably. it's a really bad idea to force it to be always fixed. some of the more common ones, maybe

Adám

10:15 AM

@Pavel Already exists. Just needs to be added to langserver.org

forest

11:17 AM

What's the general rule for a golfing answer that requires a specific version of a program to be installed, e.g. when the answer uses system() or similar?

Adám

@forest Just state so in the header.

forest

@Adám Does it count towards byte count?

Adám

@forest Any import statement does count, but the version number itself of what you use doesn't count.

forest

Then would getting rid of an ELF header by using a buffer overflow exploit on an old, known-vulnerable version of a program (e.g. ffmpeg) to execute shellcode be legal?

Adám

@forest If it actually works, then yes. Languages are defined by their implementation. This includes bugs.

forest

11:23 AM

Ah cool, thanks

flawr

1:11 PM

@Riker /r/BirdsArentReal/

Poke

3:08 PM

govavailability.info

8

you're welcome

Quintec

@Poke I just realized the 0% statistic lol

Poke

holy snapchat stock, batman

> Despite Snap's efforts to reassure investors, having a CFO resign after just eight months on the job is never a good look

Riker

3:23 PM

@flawr facts

Pavel

4:39 PM

@Adám :O

@Adám Do you know what Optima Systems is? Are they affiliated with Dyalog?

Adám

4:52 PM

@Pavel Optima uses Dyalog APL, so they are our customer. However, we often cooperate on various projects, us lending them resources, and they lending us. E.g. they maintain and develop RIDE because they have better JS manpower than we do.

Pavel

Dangit I have to reinstall dyalog again to check out the langserver

Been using TIO for so long I didn't realize it's not there anymore

AdmBorkBork

There are programming environments other than TIO?

Pavel

TIO feature-request: langserver support

Dennis

Wrong room. I won't see it here.

Also, what's a langserver?

Pavel

@Dennis It was a joke

It's a protocol for text editors to do autocompletion

So instead of creating a Sublime plugin, and Atom plugin, a VSCode extension, etc. you just write one language server

We were just talking about the APL language server

--jelly langserver when--

Dennis

5:00 PM

Heh, auto-completion for golfing languages.

J. Sallé

That'd be a doozy

Pavel

The APL langserver tabcompletes backtick-char into the appropriate APL symbol

J. Sallé

"Hey you can complete this link with any of these 89 characters"

DJMcMayhem

Just hit tab as many times as you need to

Pavel

Throw in some syntax highlighting and boom

enchanced jelly editing experience

J. Sallé

5:03 PM

Now I'm not sure what would be worse to implement

Well syntax highlighting wouldn't be too hard to get working, I guess. Problem would be to get it working correctly

Pavel

I think syntax highlighting would be simple though

Strings, numbers, other nilads/monads/dyads, maybe a special color for the two-byte math builtins

DJMcMayhem

Syntax highlighting would be incredibly ridiculously easy for one of my languages, and incredibly ridiculously hard for my other one

J. Sallé

Yeah it wouldn't be that hard to check whether the glyph is a nilad, monad or dyad

Pavel

@Dennis Imagine typing out a Jelly string and tab completing it to its compressed form

J. Sallé

@DJMcMayhem which one is brain-flak?

Dennis

5:05 PM

OK, that would be pretty neat.

Same thing for integer literals.

DJMcMayhem

@J.Sallé The easy one

J. Sallé

Yeah that's what I assumed hahahah

Skidsdev

@DJMcMayhem what's the hard one?

Pavel

The other one is V

DJMcMayhem

There are 8 distinct atoms and 1 way to do comments. The atoms are not contextual at all, they mean the same thing no matter what

V would be incredibly hard because I wouldn't even know where to begin categorizing commands.

And even something simple like "string literal" would be really hard since there are a lot of minor variations and contextual things

lirtosiast

5:21 PM

@Dennis That would be almost impossible by definition

Pavel

15 mins ago, by Pavel

@Dennis Imagine typing out a Jelly string and tab completing it to its compressed form

lirtosiast

That would be nice

New Main Posts

2

Q: Matrix rotation sort

Lets define a non-empty, unsorted and finite matrix with unique numbers as follow: $$N = \begin{Bmatrix} 4&5&7\\1&3&6 \end{Bmatrix}$$ Lets define 4 matrix moves as: ↑* (up): Moves a column up ↓* (down): Moves a column down →* (right): Moves a row to the right ←* (left): Moves a row to the le...

code-golf matrix sorting

Neil

5:45 PM

@NewMainPosts I'm pretty sure this is a duplicate...

14

Q: Rubik-sorting a matrix (a.k.a. the torus puzzle)

The idea for this code-challenge is simple: given a matrix of integers, let's sort it by applying Rubik-style movements. This means that you can select a single row or column and rotate its elements in any direction: [1, 3, 2, 4] => [3, 2, 4, 1] (rotate left for rows/up for columns) [1, 3, 2, 4...

code-challenge array-manipulation sorting rubiks-cube

Adám

@Neil That one rotates one row or column, this one moves one row or one column. Quite different, I think.

Neil

@Adám well, the moves look the same to me, but one's code-challenge while the other's code-golf so they are sufficiently different after all

(also, nice idea on the modified UTF answer of replacing the byte sequence with another sequence with the same validity)

Adám

6:02 PM

@Neil Thanks. However, isn't it a bit cheaty to just try to decode using a built-in?

Neil

@Adám ... isn't that what built-ins are for?

Adám

@Neil Sure, but it would be more interesting to actually write the code to check.

Poke

@Adám we're talking about code golf, right?

built-ins are usually golfier than writing the implementation yourself

AdmBorkBork

.NET would (sometimes) disagree with you

Pavel

@Poke Mathematica

Adám

6:15 PM

@Poke Certainly so in APL. I can't think of a single built-in that I can write shorter myself.

J. Sallé

@Adám new built-ins though, that's your thing :p

Pavel

@Adám Anything that starts with ⎕USING

Adám

@Pavel That's not built in. That's calling out to .NET

@J.Sallé Sure, but all are shorter than the currently necessary expression.

J. Sallé

Exactly why I mentioned it hahahah

Pavel

7:08 PM

@Adám Is dyalog-extended like staging for future dyalog builtins

Adám

@Pavel Not in any official manner, but I do link to its TIO when I suggest things internally. NARS also functions as a kind of testing framework for advanced ideas. Some of the content in the dfns workspace has such a role too.

New Sandboxed Posts

7:28 PM

0

A: Sandbox for Proposed Challenges

Exploding Numbers Lets define a matrix of 9s as: $$ N = \begin{bmatrix} 9&9&9\\9&9&9\\9&9&9 \end{bmatrix} $$ Lets define an exploding number as a number at position \$(x,y)\$ that can be decomposed in equal parts greater than 0 between all its adjacent neighbors (including itself). From the pr...

Pavel

I need every college and university ever to understand that if I haven't started my application with them that does not mean I would like to be reminded that there are only 6 hours left before the deadline

Adám

@DJMcMayhem @Dennis @Mego Can you grant chat.stackexchange.com/users/390618/user456082 write access to chat.stackexchange.com/rooms/52405/the-apl-orchard?

DJMcMayhem

7:49 PM

Done

Adám

8:00 PM

@DJMcMayhem Thanks!

Veskah

10:48 PM

Hmm, I wonder if Hungry Mouse on a Torus would add anything substantial. There would certainly be more empty boards

New Sandboxed Posts

11:38 PM

0

A: Sandbox for Proposed Challenges

Turtle with a portal gun The turtle wants to move along the grid to get to his food. He wants to know how many moves it will take for him to get there. As well since he is slow he has teleporters set up around his domain that he will utilize if it shortens his path. Or avoid them if it lengthe...

Transcript for