The Nineteenth Byte

Zacharý

12:01 AM

For typing it all out?

quartata

um, yes

adding an entry to a dictionary is way easier than adding a new branch

i mean just in pure keystrokes at minimum

One colon overhead vs elif, scrolling through your massive if else chain aside

Pavel

@quartata We're not talking about if/elif/else, we mean actual switch/case syntax like in C that is implemented in Coconut and perhaps should be in Python

quartata

Right but switch/case mostly just obscures the code smell

totallyhuman

I honestly don't see the appeal of switch (unless it has extra cool Haskell-like pattern matching)

plus, with python's clean if statement syntax, emulating switch doesn't lead to inelegant code

Pavel

12:18 AM

@totallyhuman That's the thing, Coconut does do pattern matching

totallyhuman

yeah but most arguments for switch don't talk about pattern matching and usually analogize to c-style languages

Οurous

I'm so used to languages with pattern matching that I'm surprised when I use one without it.

totallyhuman

also coconut seems to be following f# and not Haskell :(

Pavel

@totallyhuman I think that makes sense though. Both F# and Coconut were designed to essentially be impure, functional add-ons to an imperitive language.

Also, I like F#, so shush

totallyhuman

:P

not really against f#, it's just if coconut were to follow Haskell, it would be a combination of two of my favourite languages

Zacharý

12:26 AM

Wait, so what exactly is Coconut?

Pavel

@Zacharý Functional language that transpiles to Python

Also, all valid Python is valid Coconut. It's just an extension.

See coconut-lang.org

Zacharý

So ... sorta like (don't kill me for this comparison) ... obective-C ?

Pavel

@Zacharý Can't say. Don't know O-C

totallyhuman

...but objective-c isn't functional so I don't see where you're coming from

3

Zacharý

I mean the extension/superset part.

Pavel

12:30 AM

Oh, sure, you could say that.

Except Coconut compiles to Python directly

Οurous

So its speed is limited by the deficiencies of the Coconut compiler and the Python implementation it's running on?

Zacharý

So it transpiles? So it's like an old version of C++?

Dennis

@Pavel At least in theory.

Zacharý

Bug or feature?

Οurous

"All good Python is valid Coconut"?

Zacharý

12:34 AM

Bug in the tokenizer?

Pavel

Definitely bug

@Zacharý yes

Οurous

When I add spaces it takes 2 seconds to run that code :|

Zacharý

So the tokenizer has gone a bit COCONUTS ? (I'll stop)

Pavel

@Οurous I wouldn't say that. Coconut is still very close to Python, there's not much for the transpiler to fuck up

Zacharý

But somehow it manages to fuck up what it can

Pavel

12:37 AM

The transpiler is very slow, admittedly.

Zacharý

Yep

totallyhuman

hey can anybody link me to tacoscripts?

Zacharý

print(1) takes 2.74 seconds

totallyhuman

wowee

Οurous

Comparing the speed of transpilation + execution in Coconut / execution in Batch for the same programatic time complexity would be an interesting graph

Pavel

12:42 AM

@Zacharý ahem, 1 |> print please.

Zacharý

3.404 s

(Which seems non-consistent)

totallyhuman

lol

nvm got tacoscripts

Οurous

the "Hello, World!" is faster than Clojure's one

Dennis

Which is a lot faster than F#'s.

Οurous

I wonder what the fastest "Hello, World!" (including compile time) on TIO is.

Probably bash/dc/bc or similar

Pavel

12:53 AM

Slowest is Ceylon iirc

totallyhuman

kotlin is also up there

Pavel

Yeah, but Ceylon was > 15s last I checked

Οurous

I'm getting only 9s for Ceylon

CPU share: 123.32 % lol

... what? Ceylon reliably gets more than 100% CPU share. Am I misunderstanding what that is?

Dennis

100 % = 1 core

Zacharý

I'm getting 8.379 for Ceylon

Οurous

12:57 AM

ahh that makes sense then

totallyhuman

protip: use ataco's highlight.js userscript instead of the prettify one

it's so much better o0

HyperNeutrino

programming language name idea: subjective-c

totallyhuman

ಠ_ಠ

Zacharý

Wouldn't that be like C-- (aka a pointless name)

Οurous

o i c?

Zacharý

12:58 AM

@Οurous Eh?

Οurous

Oh I See -> o i c / O-iC -> O-C -> Objective C

ಠ_ಠ

ಠ_ಠ

ಠbjective-c?

Nಠ

1:02 AM

inb4 trash+kick (let's stop :P)

Pavel

@Οurous Looks like new servers are helping out

Zacharý

What the hell, sexefunge is a thing?

Septefunge?

Pavel

@Zacharý all those extra funges were added at once since a certain befunge implementation had flags for extra dimensions

Zacharý

How does separation of dimensions work in that? I know Trefunge uses NL and LF.

Pavel

@Zacharý No clue, check docs?

Zacharý

1:05 AM

Doesn't look like there are docs

Οurous

Oh I love languages that you learn by looking at the compiler / standard library source!

Zacharý

What do they use, CONTROL CHARACTERS?

New Sandboxed Posts

0

A: Sandbox for Proposed Challenges

Smallest/Best Compression Goal: Implement a lossless string compression algorithm. Scoring: code length * total byte count of all compressed strings Test Cases "The quick brown fox jumps over the lazy dog" "Smile miles mile" "A man, a plan, a canal. Panama." "asdfasdfasdfasdfasdf" "123456789...

Zacharý

@Dennis, you copied the source, you have any idea?

Dennis

I don't even know Befunge...

Zacharý

1:08 AM

>_<

I have no clue, it's probably TAB or something

HyperNeutrino

RLE /s

Christopher 2EZ 4RTZ

ello

Zacharý

Anybody here a wizard with Befunge?

Christopher 2EZ 4RTZ

nobody is good at befunge

except dennis most likely

Zacharý

> I don't even know Befunge...

Holdenson or someone?

Οurous

1:17 AM

@Zacharý From now on, Befunge will always sound like a magical disease that only wizards get. At least to me.

Christopher 2EZ 4RTZ

I cast Befunge! It did 15 bytes of damage!

Zacharý

ಠ_ಠ

Yeah, I give up on looking at the code for Quadrefunge, even though that would be some MASSIVE polyglotting capabilities

\f is the escape character for line feed, right?

Οurous

I think it's \r

oh wait no isn't line feed == newline == \n?

Zacharý

For that matter, how does moving between the 4th+ dimension work??

UNICODE?

Οurous

Aren't there still \f = form feed, \v = vert. tab, \b = backspace, \e = escape, \a = bell to use?

Christopher 2EZ 4RTZ

1:27 AM

We are in the third dimension being pushed through the fourth dimesion

Zacharý

I have no clue how these funges work!

Anyone want to dig through the source code to find it?

Οurous

I'll give it a shot

Zacharý

Okay, it's the quadrefunge/quintefunge/sexefunge/septefunge thing

HyperNeutrino

hexafungony

Οurous

How do you swap dimensions in normal funges?

Zacharý

1:39 AM

> means right, ^ means up, v means down, < means left

You don't really swap dimensions, IIRC

Οurous

U = up?
D = down?
P = pop-if

Those three are only active in 3+ dimensions

U / D are counterparts

P is a conditional pop of something from something

Zacharý

Yeah, Trefunge is already a thing

I'm wondering what the 4D, 5D, 6D, and 7Ds are?

Οurous

There are no further commands, but the behaviour changes. It seems to introduce something involving a higher-dimensional delta into all directional commands.

Zacharý

THat's weird

Οurous

I'm going to keep looking because this is fascinating.

Zacharý

1:52 AM

Yeah

Seems fascinating, but I should keep studying for NACLO. I'll look further into it on Thursday

Zacharý

2:05 AM

*Friday

Zacharý

2:37 AM

@all, do you think PPCG can get its own 404 page?

Dennis

Before it gets a design? Highly unlikely.

Zacharý

2:51 AM

Okay

Hey, ðproduction's here

Pavel

ಠ_ಠ

Zacharý

PavL

Pavel

3:07 AM

What happens if a cronjob is supposed to go off while your computer is off? Will it just activate the next time your computer is on?

Οurous

Is the question specific to cron, or the idea of scheduled jobs in general?

Pavel

@Οurous Specific to cron

Οurous

I know Windows' task scheduler allows options for ignoring that task run, resuming when it turns on again, or waiting for the next scheduled interval. I imagine cron, and most other schedulers, will have a similarly logical set of behaviours you can use.

ETHproductions

@Zacharý now that would be a hell of a username

Zacharý

If ý would be a valid username, I would most definitely use ý

ETHproductions

3:11 AM

I have named my JS subset JSLambda, although now I think JSFudge would be a better name, seeing as the challenge this came from was named "fibfudge" (and it bears resemblance to the name of the most famous JS subset)

brb, renaming repo

Zacharý

#ðλ

Οurous

@Pavel serverfault.com/questions/52335/… Apparently, sometimes

Pavel

Hmm. My package manager seems to have died on me, and can't be killd. I wonder how bad of an idea kill -9 would be.

Also, I'm not quite sure how this happened, but for some reason I can only update packages with pkcon update. dnf update always says dependencies resolved, nothing to do.

Dennis

Removing PackageKit is one of the first things I do with a fresh install. Solves most problems with dnf.

Pavel

@Dennis I've considered that but feared breaking everything. Glad that's not the case.

Hold on...

Removing dependent packages:
 Discord                                      x86_64                       1:0.0.4-1                               @@commandline                       138 M
 Discord-installer                            x86_64                       1.3.4-1.fc27                            @tcg-discord                        3.9 k
 flatpak                                      x86_64                       0.10.2.1-1.fc27                         @updates                            4.4 M

@Dennis Wait how, stuff depends on it

Dennis

3:24 AM

Hm, I haven't used Fedora Desktop in a while. plasma-pk-updates is probably some GUI updater; I don't use that stuff. No clue what Discord is.

Pavel

@Dennis Oh, Discord is a thing I installed manually, it's kinda like Skype. I'm more intresting in everything else.

Dennis

And a video chat client absolutely requires a package management service. Makes sense.

Pavel

@Dennis No, see, Discord cease-and-desisted to get the Discord copr repo taken down for hosting their tarball. The package I installed is Discord-isntaller, an awful hack that automatically downloads the tarball and builds an rpm out of it every time there's an update

DLosc

@Pavel I'm sure you've gotten this before, but: Obligatory XKCD X^D

Pavel

Although I just discovered there are packages on the OpenSUSE build server, so eh, don't need packagekit

Dennis

3:32 AM

@Pavel Oh noes, somebody provided a convenient way to install their software. How dare they!

6

MD XF

3:55 AM

Does anyone know wtf is wrong with this

Οurous

@MDXF you forgot a { after your if

MD XF

Omg come on I've been puzzling over this for twenty minutes

Οurous

C/C++ compilers do have some of the most unhelpful errors and warnings

MD XF

I sincerely agree

Dennis

Even for C, error: initializer element is not constant is pretty unhelpful for a missing {.

@MDXF Could have taken a few seconds. tio.run/##VY5BT4MwGIbv/…

MD XF

4:02 AM

@Dennis I didn't realize how useful -Wall got with the latest GCC

Pavel

Goddammit the discord opensuse package is broken. Mind you, it's for Fedora 27, it's just on the opensuse build server.

I've tried just the tarball, the snap package, the suse package, and they all don't work properly.

Back to packagekit and Discord-installer

Anonymous

@MDXF -Wall has always been useful

Anonymous

Bonus points for -Wall -Wextra -Werror -pedantic

MD XF

-pedantic is totally stupid.

"ooh let's enforce fake rules that couldn't possibly have the slightest impact on behavior :D"

Anonymous

@MDXF Every word of that was incorrect

Anonymous

4:10 AM

-pedantic enforces strict ISO compliance - it helps enforce portability

MD XF

Yes but what modern compilers require strict ISO compliance?

Anonymous

A single modern compiler doesn't. Supporting multiple modern compilers does.

Anonymous

Not everybody uses GCC

MD XF

Yes, but most people use compilers that don't require all/any the things -pedantic enforces

Zacharý

@Mego <disapproval face here>

Οurous

4:12 AM

@MDXF It isn't for people who provide binaries produced with different compilers, it's for people who distribute the source and don't know what compiler the other guy has

Anonymous

@Zacharý Clang is actually better than GCC in some ways

MD XF

@Mego What ways are those?

Zacharý

Reasons

GCC > Clang. GCC (IIRC) will eventually support D, so GCC will make my life complete.

Anonymous

@MDXF Clang plays nicer with debugging and linting tools

MD XF

@Mego ooh what's clang's debugger?

Because gdb isn't really that useful

Anonymous

4:14 AM

@MDXF lldb

MD XF

Because LLVM, I'm guessing?

Anonymous

Also Clang is reentrant and can compile code in a multithreaded manner, unlike GCC

MD XF

That's actually really interesting

Anonymous

Also Clang's error messages don't suck

MD XF

CMP: Adding "solve cube" builtin to Cubically. What Unicode character should I use for the command?

@Mego Ok brb switching to clang

Zacharý

4:15 AM

Ř, for a friend

Anonymous

Previous C code, compiled in clang

Anonymous

And to top it all off, clang is usually faster :P

MD XF

error: extraneous closing brace WHY does GCC not have this ?!

~~It didn't warn me once~~ wait no it did. It didn't warn me helpfully I had a random closing brace

Pavel

@MDXF This is your fault for not using -Wall

@Mego Hold on, why do people use gcc then?

Zacharý

We need to build with Wall

Anonymous

4:19 AM

@Pavel Because clang isn't as widespread/well-known as GCC

5

Zacharý

(please don't kick me, O powerful penguin)

Anonymous

GCC has the backing of the FSF, while LLVM and Clang are independent projects

MD XF

Could GCC technically support an LLVM backend?

Οurous

@MDXF The ▦ - which looks like a uniformly colored rubiks cube side

Anonymous

@MDXF Maybe, but not easily. GCC is largely monolithic, so it's hard to take chunks out and replace them

MD XF

4:21 AM

@Οurous I was thinking about that, it's kinda annoying that it's a 4x4x4 though because this builtin can only do 3x3x3, but if I can't figure out something better that's what I'm gonna go for

Pavel

@Mego I meant that in the sense of "What does Clang do worse than gcc so I have something to think about before switching"

MD XF

While we're talking about C, I think everyone should see this: github.com/shinh/elvm It compiles C to esolangs like Brainfuck, Befunge, Piet...

Anonymous

@Pavel Clang doesn't have the backing of the FSF, so they have less money, and thus it takes them longer to implement new standards. Also they don't have the weight to throw around to influence new standards as much.

Pavel

Well, the last standard was C11, so I hope everyone's got that by now. I think I like clang

Anonymous

@Pavel C++17 was just last year

Zacharý

4:25 AM

But not everyone's got C99

Anonymous

Also MSVC still doesn't support all of C99

Οurous

@MDXF Technically it's an NxN(eg: DejaVu has it as 5x5, my browser has it as 4x4, ...)

Zacharý

'cause dynamicallyinited stuff

MD XF

@Οurous Ah interesting. I'll use it then

Pavel

@Mego D: Wikipedia page for C11 still says it's current

Anonymous

4:26 AM

@Pavel For C, C11 was the latest standard. For C++, it was C++17, and they're working on a new one already.

Anonymous

@MDXF ▣

Pavel

Ah

Anonymous

C++20 will include concepts, which I'm excited about

Οurous

@Mego Which concepts in particular are exciting?

Pavel

@Mego Are those the metaclasses?

Anonymous

4:29 AM

@Pavel Yep

Οurous

Oh. Like concepts. As a thing like classes?

Anonymous

Also possibly coroutines, transactional memory, and reflection

Anonymous

Concepts are C++'s way of doing pseudo-duck-typing

Anonymous

Concepts (C++)

Concepts are an extension to C++'s templates, published as an ISO Technical Specification ISO/IEC TS 19217:2015. They are named boolean predicates on template parameters, evaluated at compile time. A concept may be associated with a template (class template, function template, or member function of a class template), in which case it serves as a constraint: it limits the set of arguments that are accepted as template parameters. The following is a declaration of the concept "EqualityComparable" from the concept-enabled C++ standard library (which is a separate ISO Technical Specification, ISO/IEC...

Anonymous

They allow specifying restraints on template args

Zacharý

4:31 AM

Didn't C++17 include variant?

Anonymous

Yep

Dennis

@MDXF ELVM IR would make a nice addition to TIO.

Οurous

Oh neat. So templates are getting closer to a proper generic typing system?

Anonymous

@Οurous Yep! It's exciting

MD XF

@Dennis I agree, would you mind adding it then?

Zacharý

4:32 AM

Maybe they'll fix complex then :p

MD XF

What's wrong with complex?

Pavel

@Mego I'm just glad we can now have actual interfaces

Zacharý

complex of complex

no

Pavel

MSVC has interface as an extension, but extensions aren't nice.

IDK about other compilers

Dennis

@MDXF I'll see what I can do.

Zacharý

4:50 AM

My challenge is actually popular, woohoo!

MD XF

@Zacharý 404 challenge?

Zacharý

Yeah

MD XF

I like how different it is from other polyglot challenges, I've never had to write a polyglot that intentionally breaks in certain languages

Zacharý

Yeah, didn't want a bunch of 404s and print(404)s

MD XF

Well you got a print(404) :P

Zacharý

4:53 AM

I mean for the final answer :P

MD XF

@Zacharý ... it is

Zacharý

... >_<

Creative though

MD XF

Yeah it took forever to slowly get it to stop borking in each language

I also thought of making one that just ended up with 404 and I had fourteen langs ready to go but I already had two answers and one is going to win (I strongly hope)

Zacharý

I'm expecting A FlakFlakFlakFlakFlak answer

Sorry

I meant FungeFungeFungeFungeFunge

The one I got beat by in the Hello World in Multiple Languages.

MD XF

20

A: Hello World in Multiple Languages

23 Befunges, 713 bytes The only language I really know is Befunge, so instead of multiple languages, I've just gone with multiple implementations of Befunge. I believe this is still valid under PPCG rules, which consider the language to be defined by its implementation. 8023/# !-1401p680p88+79*...

Like ^ ?

Zacharý

4:59 AM

Exactly like that

That's the one I'm talking about

55 upvotes though!!

MD XF

On what now?

Zacharý

That challenge, I got 55 upvotes for my answer. It's under the flurry of Add++

Pavel

Why is it that everyone and their dog went and made a Befunge interpreter?

MD XF

Ah that one

@Pavel Everyone and their dog, cat, grandma, fish, and third cousin went and made a brainfuck interpreter so I guess Befunge should count itself lucky

Zacharý

I have no clue. Befunge 97 is a thing, and so is Septefunge (!))

MD XF

5:02 AM

I just added a cube solver to Cubically and its code coverage dropped from 92% to 35% D:

Pavel

@MDXF Well brainfuck is easy. The version TIO uses is under 30 lines of code.

Zacharý

What is "code coverage"

MD XF

@Zacharý percentage of source lines your code actually uses

Zacharý

@Pavel And yet it still has like a million interpretations as well. (Wrapping/Nonwrapping Infinite/Noninfinite 0 -1/0/<do-nothing> EOF)

MD XF

int main(void)
{
    puts("Hello, World!");
    return 0;
    puts("Haha nope");
}

The code coverage for ^ would be 66%. Function/variable declarations and random stuff like { are skipped; only the real code counts

Zacharý

5:04 AM

Ah, I see

Pavel

@MDXF I think I get it less now

MD XF

Code coverage

In computer science, test coverage is a measure used to describe the degree to which the source code of a program is executed when a particular test suite runs. A program with high test coverage, measured as a percentage, has had more of its source code executed during testing which suggests it has a lower chance of containing undetected software bugs compared to a program with low test coverage. Many different metrics can be used to calculate test coverage; some of the most basic are the percentage of program subroutines and the percentage of program statements called during execution of the test...

how about now

Zacharý

He's disregarding the function header

Pavel

@MDXF Right, coverage only makes sense in the context of tests.

A program by itslef doesn't have code coverage

MD XF

Well technically it does; it's just not as useful

Zacharý

5:10 AM

>_<

MD XF

Bam back to 89%, gotta love .yml configurations

Oh no @dylnan you're back on PPCG that means I'm not going to be winning the polyglot challenge for long D:

dylnan

@MDXF haha nah I probably won't be on for much longer

I mostly came on to take out Brainbash and fix Numberwang

MD XF

chants tauntingly I'm only two langs ahead....

dylnan

I might get one or two more in ;)

I think if I had left out Jelly, M, Pyon and Proton I would have been better off. Those four restrict the way I can start lines a lot

Zacharý

If any of you gets to 404 (VERY doubtful), you win for sure

dylnan

5:14 AM

Haha that would be great

What about 44

Zacharý

We should probably have a custom site by 404

dylnan

Because 0 is basically the empty string

Zacharý

But by then, the behemoth would've passed 404.

MD XF

@dylnan I intentionally didn't use languages like that

I realized Py(th)on was easy to add around #25

dylnan

prudent

I didn't think ahead at all

MD XF

5:15 AM

@Zacharý will you bounty? (If it's not me I will for sure)

@dylnan Well I thought ahead more in the one that ended up with print(404) :P

dylnan

True. I bet you could take that one pretty far

with fewer bytes anyway

Zacharý

If you get a polyglot that has MY, D, APL, and Fortress: I will give you a ton of rep

MD XF

I've tried adding more, everything keeps succeeding at print(404); and I'm not eloquent enough for this joke

Zacharý

(Using unicode codepoint values for MY, APL + Fortress)

(This is unofficial, of course. Don't want those behemoth nutjobs going on this)

MD XF

@Zacharý I'll go for it when I'm done decimating the "solve rubik's cube" challenge with Cubically's new builtin

Zacharý

5:21 AM

Two of those languages have weird-ass code pages, D is bad with polyglotting, and Fortress is ... incomplete to say the least. (RIP Fortress, you would've been the only language I program in for non-golfing purposes [maybe except APL])

MD XF

Polyglotting with Cubically and Implicit (and soon Integral) is hard because they use the same parser and syntax

Zacharý

Polyglot D into a bunch of languages, and I might just give rep.

MD XF

Bam. I hate myself.

But neither the language nor the builtin were made for the challenge; it's just a language that can (well, that will be able to) do virtually anything that has to do with a Rubik's Cube...

Zacharý

That's going to have some downvotes

MD XF

I find that likely. But if every time Cubically destroys a rubiks-cube challenge it gets downvotes, well, I'm not just going to stop winning.

Jelly wins pretty much every code-golf challenge because it's designed for code-golf; Cubically wins pretty much every rubiks-cube challenge because it's designed for rubiks-cubes. But Jelly doesn't get downvoted very often, so I hope Cubically doesn't. It hasn't yet.

Zacharý

5:26 AM

I thought Cubically was just ON a Rubix cube

MD XF

And it's not as if the ▦ builtin just prints moves to solve the cube; it works like an eval statement and sticks the moves to solve the cube in the source code. It only uses an optimal algorithm so that the interpreter has less to do

@Zacharý Nah Cubically just does anything that has to do with Rubik's Cubes

Zacharý

...

MD XF

¯\_(ツ)_/¯ just don't downvote pls @else, feel free to not upvote though

Zacharý

I won't

Esolanging Fruit

5:42 AM

Random question: Consider BF with two different loop types (let's call them [] and {}) that function exactly the same but can be interleaved (like [{]}). You can't nest loops of one type inside each other (no [[]], no [{[]}], but yes {[][}][]. Is this TC?

Zacharý

I dunno

dylnan

Isn't that a superset of BF? If so it is TC

Leaky Nun

@dylnan BF allows nesting

dylnan

Oh nevermind

@LeakyNun thanks I didn't catch that at first

Οurous

Can you do anything with a nested loop that you can't do with a single loop and some preparation code?

Or rather, with one-deep nested loops.

Zacharý

6:05 AM

Hello unihedron, or more appropriately polygon.

Unihedron

A unihedron is not a shape.

Zacharý

I extrapolated things, okay?

Esolanging Fruit

@Zacharý Wouldn't that be a Dihedron?

Looks like a dihedron isn't a completely trivial structure in non-Euclidean space

Zacharý

Ughh, forgot about that. I just can't seem to sleep today. So I'm doing this instead until I pass out

Esolanging Fruit

or something

@Zacharý I think the correct approach is "sleep until you can do math", not "do math until you can sleep"

Unihedron

6:19 AM

A regular polyhedron with one regular enclosing shape is called a henahedron, it exists in other geometric systems.

Zacharý

@EsolangingFruit My body should pass out in a few minutes, hopefully

DLosc

7:02 AM

@Οurous I'm curious to know the answer to this question--or, more generally, how many levels of loop nesting you need in BF for it to be Turing-complete. If the answer is suitably small, it would make a Turing-completeness proof for Acc!! much easier.

aditsu

9:23 AM

hahaha, I just found this article via imgur.com/gallery/Gnd30Nm

Adám

Any ideas for this?

Martin Ender

@Adám I disagree with that comment, the challenge makes more sense if your solution can assume that it's given a function expecting two integers and returning one. That said, you could probably make this clearer by talking about a generic blackbox function f right away and then just giving some example functions as test cases later on.

Let me see if I can find a similar challenge as an example.

Ah yeah, this is the one I was thinking of: codegolf.stackexchange.com/q/150186/8478

It's fine if languages without higher-order functions can't participate.

Adám

@MartinEnder Great feedback. I'll piggyback that.

@MartinEnder How should I "picture" the function? ``f: ℤ,ℤ → ℤ` ?

Martin Ender

yeah, that works

normally with an x instead of a comma though

or rather an ×

Adám

@MartinEnder OK, but why, though? Looks like multiplication to me.

Martin Ender

9:38 AM

it's a cartesian product

ℤ × ℤ is the set of all ordered pairs of integers

Adám

@MartinEnder {0,0} is not in the domain of ÷ or exponentiation (according to many).

Martin Ender

then either ÷ is not a valid input for this challenge, or you'll need to make the spec a bit more complicated and say that you can assume that all intermediate function applications will be in the domain of the given function

Adám

I'll add that you can assume that the diagonal values are in the domain of the function.

Martin Ender

it seems easier to restrict inputs to total functions on ℤ^2 though

Adám

@MartinEnder Why? Division and exponentiation are pretty basic. As long as the input will be strictly positive, there's no issue.

Martin Ender

9:42 AM

because it doesn't affect the actual challenge but simplifies the spec

Adám

@MartinEnder Either I have to specify that the functions will be total, or that the matrices will always be in the domain. Either way, I have to add another spec rule.

Martin Ender

but you can also make it f : N+ × N+ → N+ (superscript +) and allow people to assume positive integer matrices

Adám

@MartinEnder Hm, yeah, that doesn't detract from the challenge, and it allows n-root-m.

Martin Ender

division still doesn't work though unless you make it round up

Shaggy

o/ 'Lo, all

Adám

9:45 AM

@MartinEnder How about ℕ⁺ × ℕ⁺ → ℤ ?

@Shaggy Just in time for the discussion of your comment. Nice!

Martin Ender

doesn't help, because you need to feed the result to the next function call

Adám

@MartinEnder Oh, right. D'oh.

Hm, I think I'll just go with a rule that all values and intermediary values will be in the function's domain.

Shaggy

@Adám Timing! (I assume you mean the one on your sandboxed challenge?)

Adám

@Shaggy Yes.

Shaggy

My initial reading of the challenge implied that you had a set list of functions in mind that solutions must be able to support; your reply clarified that.

Adám

9:55 AM

@MartinEnder @Shaggy Ready to post?

Martin Ender

"A matrix with at least two rows and two columns" -> "An integer matrix..."

Adám

@MartinEnder No, see examples.

Martin Ender

That example is not f: ℤ × ℤ → ℤ

and requiring non-integer types adds nothing to the challenge, imho, it just makes it a pain for languages that strictly differentiate between integers and floating-point types

Adám

@MartinEnder Input can be all-float for convenience, no?

Martin Ender

I guess. still seems unnecessary

Shaggy

9:58 AM

The addition of the f(x,y) example and specific mention of black box functions definitely makes it clearer.

Martin Ender

I'd also find the explanation clearer if you skipped straight from the regular trace example with + to using just f in your notation instead of using other operators. that's probably due to your APL background, but in most other languages, you wouldn't actually insert operators and worry about default associativity.

you'd just transform it to f(f(a,b),c) or f(a,f(b,c)).

Transcript for