The Nineteenth Byte

thejonymyster

00:21

@emanresuA is there supposed to be an equals sign there

emanresu A

Yes

NoHaxJustRadvylf

Hey quick question for anyone here with experience with servers, you can put different lengths of servers in the same rack, right? E.g., if one's 36" and one's 33" (made those numbers up not sure how accurate they are).

thejonymyster

@emanresuA where would it go then :P

emanresu A

v_1 ... v_l = I_i

thejonymyster

thank u

OldSandboxPosts

01:00

Sandbox posts last active a week ago: Is this a squashed series? [code-golf] [string] [parsing] [number] [decision-problem], Solve a Card Suit Puzzle, Interleavable programs (WIP) [code-challenge] [source-layout]

New Posts

0

Q: Golf Github's Pager Logic

Background When searches return many pages of results, Github avoids cluttering their UI by eliding page links. Instead, their UI let's you select: Pages at the beginning of the results. Pages near the current page. Pages at the end of the results. In this way, no more than 11 page links are e...

code-golf

emanresu A

02:13

I wonder if there's a way to encode Piet more golfily

NoHaxJustRadvylf

There most definitely is

It only uses a certain number of colors, and the way it's designed with the grouping of similar colors makes it very much not golfing oriented

Plus there's more unusual encodings for 2d languages, like mine, which are more compact

Bubbler

I guess many Piet programs benefit from omitting the description of black cells entirely

e.g. writing it as "walk description", like RabcDdef

Seggan

@emanresuA yeah, use asciipiet with fractional encodings

emanresu A

02:54

CMQ: Is there a way to overload the find function in Python?

*index

Never mind, I'm anidito

pxeger

05:44

@NoHaxJustRadvylf Yes. You can put a 30" depth server in a 36" rack, but not a 36" server in a 30" rack. (And the width has to be exactly the same. The standard is 19")

Sandbox Posts

07:41

0

A: Sandbox for Proposed Challenges

Alphanumerics but no Alphanumerics We define the alphanumeric bytes as qwertyuiopasdfghjklzxcvbnmQWERTYUIOPASDFGHJKLZXCVBNM1234567890 (i.e. All letters, both cases, and numbers). The order does not matter and I typed it in that order to be lazy. Your task is to write a program that prints the byt...

Nobody Needs Names

And @emanresuA Pride Month is over

@SandboxPosts any suggestions?

pxeger

Opinions on whether codegolf.meta.stackexchange.com/a/22383 is a dupe of codegolf.stackexchange.com/q/217690?

Nobody Needs Names

my opinion is yes

and imo the title's spelled wrong

I just did XXXXXX in Minecraft and it's blablablabla

if I just said that will it get moved immediately or after the conversation or somewhere in between? To off-topic or to CGCC-gaming?

oh and what's the lang for tmr's (today's depending on time zone) lyal?

lyal is too similar to @lyxal

Neil

08:32

@NoHaxJustRadvylf as I recall you can get them with adjustable mounting kits

pxeger

rack rails tend to be inexplicably expensive though

Neil

fortunately that wasn't my problem

pxeger

08:59

Does Rust have a better way of writing this:

let definitely_x = match maybe_x {
    Some(x) => { x }
    None => { return }
};

I could use if let, but then my code gets unnecessarily nested if I'm checking several Options

And the function returns (), not an Option, so I can't use ?

Bubbler

You can write an inner function that returns Option<()>

and then discard its return value

New Posts

09:16

0

Q: Build a file/directory tree

Task In this challenge, your task is to write a program or function which takes an array of paths with an additional boolean indicating it is a file or directory and outputs a file/directory tree in any reasonable format. Remarks A directory can but should not end with a forward slash ("/") A pa...

code-golf string tree-traversal

ais523

@emanresuA Jelly, 6 bytes

(relies on the solutions being integers, if you change : to ÷ it works for non-integers too but then the output will always be a float)

@NoHaxJustRadvylf stderr can be a file, it's common when running long/complex scripts to just set up stderr to be a file directly as an easy way of logging the errors

pxeger

@ais523 I think he meant "a system in which /proc/self/fd/2 does not always refer to standard error", e.g. if /proc is not procfs, which is treated specially.

ais523

/proc actually isn't treated specially even in Linux

ah right, now you've edited i don't need to explain procfs

but I used to build containers manually quite frequently, and those will contain neither /dev nor /proc unless you explicitly mount them

pxeger

TIL that procfs is not POSIX-standardised

ais523

it's a Linux thing specifically I think

stderr appears to be write-only on TIO, which is unusual for programs being run normally, so I suspect that it is actually a file there (that's the most common case in which you get a write-only stderr)

pxeger

09:27

surely stderr is generally only readable if it points to a pty?

ais523

many people are surprised to discover that reading from stderr normally works (just like writing to stderr writes to the user's terminal rather than a redirected stdout, reading from stderr reads from the user's terminal rather than a redirected stdin, in both cases unless stderr is also redirected)

doesn't need to be a pty specfically, any sort of terminal works I think

pxeger

I don't see why any program would go out of its way to open a read-write pipe for stderr

ais523

and yes, you could intentionally open a file as read-write but it probably wouldn't do anything useful because you'd have nowhere to put the file pointer

I used to use SunOS a while back, pipes there were bidirectional by default

can you make a bidirectional pipe on Linux?

pxeger

AFAIK, TIO and ATO both just use pipes for stderr

ais523

ah yes, it's a pipe on TIO (Try it online!)

pxeger

09:30

@ais523 a pipe is just a pair of file descriptors, one for reading, and one for writing. You can use both in the same process, if you want, although normally you pass one end of the pipe to a child process, and keep the other end

ais523

@pxeger that seems to be the case on Linux; on SunOS you could write either descriptor and read back from the other

pxeger

oh, interesting

ais523

when I ported my programs from SunOS to Linux, the only thing I typically had to change was to swap the ends of a pipe if I'd connected it backwards by mistake

pxeger

FIFOs can be open read-write on Linux

ais523

unix.com/man-page/sunos/2/pipe

pxeger

09:32

You might be able to re-open a basic pipe as read-write using fcntl, I'm not sure

ais523

<Linux man 7 pipe> On some systems (but not Linux), pipes are bidirectional: data can be transmitted in both directions between the pipe ends. POSIX.1 requires only unidirectional pipes. Portable applications should avoid reliance on bidirectional pipe semantics.

so looks like there's no way to get the two-queue behaviour

I don't think it was particularly useful anyway? I only used it by mistake

pxeger

You could always just open two pipes and use both of them.

ais523

yes

except that that's almost always a bug – you need to implement the flow control manually to prevent a deadlock

pxeger

I wonder how easy it is to take a read-only FD and a write-only FD and multiplex them into one read-write FD

ais523

I think that's impossible at the FD level, and possible but fairly obscure at the stdio level, not sure though

at least with glibc

pxeger

09:37

@ais523 I suppose the advantage of two-way pipes is that you don't need to remember which of the two return FDs is which

ais523

@pxeger yes, that's why I kept connecting them backwards :-)

pxeger

@ais523 I'm pretty sure it's possible if you spawn an intermediate process to handle the reads and writes. It would only work indirectly then, though.

ais523

that SunOS man page gives zero indication of what the correct connection direction is

@pxeger ooh, wouldn't need to be a process, you could use a kernel module

pxeger

you say that like writing a kernel module is easier than writing a small userspace program

ais523

maybe not easier, but surely more efficient?

I've been considering redoing the fizzbuzz as a kernel module, but it wouldn't technically comply with the spec

pxeger

09:40

Write a fizzbuzz OS

Make a fizzbuzz FPGA

ais523

and measuring the throughput would be hard to define, as the "splice to /dev/null" approach wouldn't calculate the fizzes and buzzes at all

there was some discussion of the FPGA approach, I think the general conclusion was that FizzBuzz is possible at combinatorial speeds so it'd just be down to how much I/O bandwidth the FPGA had

you could use pretty much the same algorithm as in my answer (striping the memory differently)

pxeger

@pxeger if you actually did this, and got it to run via KVM, it might actually be valid, and work better than a normal process

ais523

a fizzbuzz OS would be almost entirely I/O

and again, it's hard to define the speed of output

speed at which you can repeatedly overwrite VGA video memory, perhaps? (displaying on the screen is a valid output method, although I'm not sure whether modern computers can even be set into VGA text mode)

pxeger

surely KVM has some kind of basic byte copying buffer

but then you're copying, so that's probably not gonna work faster

ais523

I/O is one of the slowest parts of virtualisation

if you do it the originally intended way, with in and out processor instructions and interrupt vectors, you're going to have a huge number of processor traps and virtualised context switches, which are not fast at all

those are slow even on a non-virtualised processor, though, so modern I/O is often about trying to avoid them as much as possible

(which is why motherboards are so complicated, as they have to handle not just all the I/O routing, but also identifying the triggers for the various types of I/O occurring)

PyGamer0

10:34

hello gamers

Hello, World!

11:01

0

A: "Hello, World!"

Cognate, 20 bytes Print"Hello, World!" Attempt This Online!

Sandbox Posts

0

A: Sandbox for Proposed Challenges

Convert from Greeklish to modern Greek Greeklish, a portmanteau of the words Greek and English, is a way of writing modern Greek using only ASCII characters. This informal way of writing was extensively used in older applications / web forums that did not support Unicode, and were not programmed ...

emanresu A

11:42

CMQ: Is CSS declarative or imperative?

user

Declaratice right

pxeger

declarative

user

“This is what h1s look like “ not “make h1s green then underline them”

Neil

11:58

@ais523 loopback socket? maybe a unix domain one

ais523

@Neil that wouldn't be a pipe, but you're right, it serves basically the same purpose as one

and is bidirectional

pxeger

once you've opened a connection on a socket it acts like a pipe

but there's more overhead in negotiating the connection

ais523

sockets have a bit more functionality, too

pxeger

there's also socketpair

mousetail

there's also linux file based sockets that basically have all the benefits of both

pxeger

12:01

unix domain sockets are probably the file based sockets you're thinking of

or do you mean named FIFOs?

mousetail

yes I mean domain sockets I think

Nobody Needs Names

12:45

-2

A: Sandbox for Proposed Challenges

Alphanumerics but no Alphanumerics restricted-source code-golf We define the alphanumeric bytes as qwertyuiopasdfghjklzxcvbnmQWERTYUIOPASDFGHJKLZXCVBNM1234567890 (i.e. All letters, both cases, and numbers). The order does not matter and I typed it in that order to be lazy. Your task is to write a...

While unsuited for main, this might find success as a CMC in TNB. — Adám 11 mins ago

Good. CMC.

Adám

@NobodyNeedsNames APL, 14 bytes: ∊(-∘≢↓⍳)¨':[{'

':[{' are just those three characters. Then, on each (…)¨ we find the range ⍳ which gives us 0…: and A…[ and a…{ from which we drop ↓ from the rear -∘ as many elements as there are in each of : and [ and {, i.e. 1 each.

mousetail

13:03

Time for jsfuck.com

thejonymyster

lol i was gonna try and do it with regular js :P

lyxal

I was going to do it in a half cheaty way

mousetail

How do you get letters without using letters without JSfuck shenenigens

pxeger

well you'd have to use a similar method to jsfuck

but without necessarily using jsfuck.com to generate it

mousetail

fair enough, I guess it might be more efficient if you use a few more symbols

thejonymyster

13:06

yeah

like for example theres still variable names u can use :P

mousetail

Just underscores right?

and $

thejonymyster

afaik

Adám

Accented chars and all other \w characters from all languages.

Maybe I'll start naming all my JS variables with one or more Hangul Filler character (U+3164):

thejonymyster

ok i cant sit here and do this rn why am i doing this lmao

also wow

mousetail

That's evil

I imagine you could use that to undetecable inject some malware into code

3

pxeger

13:12

→ 2 messages moved to Sandbox

Zion mycelia adamancy

@Adám Goddammit JS, why?

pxeger

Python accepts it too

it's a letter

PyGamer0

wait what... LOL

Adám

And that's why we at Dyalog don't "just" allow any Unicode letter for identifiers. (Many people have suggested that.)

mousetail

most languages do, I beleive it's a official unicode standard about what letters are alowed in identifiers that most every language follows

Zion mycelia adamancy

13:13

@pxeger Goddammit Python, why? :P

lyxal

@Zionmyceliaadamancy you think that's bad?

pxeger

@mousetail yep, unicode.org/reports/tr31

lyxal

Try it online!

Zion mycelia adamancy

@lyxal I mean, it's not great :P

lyxal

You haven't seen bad

PyGamer0

13:14

Python, 20 bytes: ㅤ=1;print(ㅤ+ㅤ)

LOL

thejonymyster

i had a terrible idea along similar lines;

reserved keywords are only reserved in the context theyre reserved in

lyxal

Turns out js accepts ඞ as a variable name

pxeger

@Adám I suppose in APL it's a more serious problem, because inserting only a variable name could cause a bigger change in the syntax

PyGamer0

@lyxal lol

pxeger

@thejonymyster like soft keywords? Those are not a terrible idea if used with care

thejonymyster

13:15

if = true; if(if){print(if)}

ais523

@NobodyNeedsNames it's actually a duplicate that did very well on main

think of it as self-validating rather than restricted-source, i.e. the interesting part is the restriction, the task is irrelevant

Adám

@pxeger Maybe, but it is still nice that it is (relatively) easy to type all valid identifiers.

PyGamer0

@thejonymyster omg imagine the amout of cursed things you can do

thejonymyster

i think itd be funny :P

ais523

experience seems to be that boring restrictions don't improve challenges even if the task is interesting, but interesting restrictions make for good challenges even if the task is boring

mousetail

13:17

What HTF does this work codegolf.stackexchange.com/a/246396/91213

thejonymyster

~~and technically golfy~~

pxeger

@Adám Pretty ironic to say that with a language so special it needs its own (part of a) Unicode block

thejonymyster

its also pretty cool that it does have that

ais523

(also I was hoping that my reply to a comment by @Adám would ping the author of the comment, but instead it pinged the user quoting it)

thejonymyster

i was unicode hunting and so surprised to see all the apl ones labelled as apl lol

pxeger

13:18

@Adám and I'm sure people who don't use a basic latin keyboard don't find them that easy to type

lyxal

@mousetail unicode normalisation

ais523

I can type almost all Jelly's identifiers, but a few of them come out incorrectly

e.g. the key sequence that for Dennis is probably ʋ comes out on my keyboard as ư

PyGamer0

@thejonymyster ato.pxeger.com/…

pxeger

@pxeger ato.pxeger.com/…

lyxal

@mousetail see also: codegolf.stackexchange.com/a/211249/78850 and codegolf.stackexchange.com/a/211250/78850

thejonymyster

13:20

@pxeger true

Zion mycelia adamancy

and codegolf.stackexchange.com/questions/207560/…

ais523

there is something wrong with the sandbox, though, if we can downvote a challenge and say that it won't do well, when it's a duplicate of a challenge at +54 (both the estimate that it wouldn't do well, and the fact that the sandbox commenters missed the dupe)

thejonymyster

@PyGamer0 i dont know the lang enough but i think thats cursed :P

pxeger

@ais523 I've only had to add the hook characters and one or two dotted ones to my layout. Dennis did a pretty good job

ais523

I have a dead key for dot, so I can dot anything pretty easily

ạȧḅḃ etc.

it's on altgr-/ (below) or altgr-shift-/ (above)

pxeger

13:23

@ais523 It did very well on main... 5½ years ago. The reason we discourage "print X without X" is because almost all interesting variations have now been done

lyxal

@Zionmyceliaadamancy dang it that's the one it was looking for :p

ais523

and I didn't even put it there myself! (although it took a while to figure out where it was on the keyboard)

@pxeger in that case, maybe we should be finding duplicates for the interesting ones rather than saying they won't do well?

pxeger

@ais523 X11 has builtin composition for most dotted letters too, just except ŀĿ for some reason

PyGamer0

@thejonymyster more cursed: ato.pxeger.com/…

ais523

@pxeger that's because the dot there isn't an accent, it's a composition of l/L with ·

thejonymyster

13:25

last time i tried to make a code sheet it was awful. there were like. braille characters. and dominos

ais523

for Catalan, where ll has a special meaning so you have to break it up as ŀl if you want two separate ls in a row

thejonymyster

@PyGamer0 classic

pxeger

@ais523 isn't the statement "they won't do well" true because of the existence of duplicates? So why bother finding the duplicates, if we know they exist? (and even if they don't exist as actual duplicates by SE's standards, we know they're overdone?)

ais523

so it's unsurprising that it acts differently from the uses of overdot/underdot as an accent (it's even in a different place on the character)

pxeger

in that case it's mainly surprising that Jelly has that character as an odd one out

ais523

13:26

@pxeger I think of it the other way round – it's clearly an interesting challenge, so if it isn't a duplicate, it will do well

Zion mycelia adamancy

@ais523 ngl, I always thought that the dot wasn't an overdot because L/l are "tall" letters :P

Plus, l with an overdot can look too much like an i

ais523

@Zionmyceliaadamancy so did I, I only stumbled across the source of the letters by chance while looking up something Jelly-unrelated

Adám

What an improvement! Compare the original to the new style.

6

pxeger

@Zionmyceliaadamancy ḷ already looks a bit too much like !

mousetail

What a terrible day to have eyes

thejonymyster

13:28

@Adám very spacious

pxeger

@Adám Somebody call Arthur Whitney right now!

Zion mycelia adamancy

@pxeger Obviously two identical characters look alike smh

ais523

I'm a bit disappointed that there isn't an ı with underdot (at least, I assume there isn't, the key sequence that should produce it doesn't do anything)

thejonymyster

i couldnt have the restraint that the unicode guys have

pxeger

For me, the character sequence which would produce "i with overdot" actually produces i with no dot: ı

ais523

13:29

you can probably craft it out of combining characters?

@pxeger for me too

Ginger

@Adám it's too early for this man

ais523

dotless I is famously a letter in Turkish

pxeger

@ais523 ı̣ ı̣

mousetail

I wonder if it would be possible to configure uglifyjs or another mimifier to generate only invisible identifiers. Wouldn't really make the code harder to reverse engineer but anyone trying will probably bleed out their eyes

ais523

you could disguise the remaining code as JSFuck?

thus causing people to use the wrong deobfuscator

thejonymyster

13:31

oml thatd be good for a cnr if not for the whole "dont be underhanded" thing

Unrelated String

@ais523 huh, i knew about the catalan interpunct but had no idea that was what ŀ is from

ais523

hmm, that almost sounds like a source-layout challenge at this point: write code which does one thing on its own, and something else if all the ASCII alphanumerics are removed

thejonymyster

ooo

pxeger

@Adám Here's a good use for font ligatures: you can distinguish =�> from => (where � is replaced with an invisible character, but I made it visible for clarity)

ato.pxeger.com/…

Adám

Uhh…

ais523

13:32

some languages could even do a three-way: a punctuation-only program, an alphanumeric-only program, and a program that's the two interleaved

pxeger

@ais523 you could allow answers to define their own subsets of characters to remove, most subsets wins (probably too easy to game though)

ais523

@pxeger that would drive me crazy trying to cheese it with A Pear Tree :-D

I only need 2-char subsets to get the checksums working, but am not sure how many disjoint subsets are even simultaneously TC there

pxeger

Why is A Pear Tree called A Pear Tree?

ais523

it was meant to be a one-off joke

pxeger

that sounds like such a ridiculous question out of context

Ginger

13:35

Welcome to Tautology Club!

thejonymyster

@pxeger pxeger get out of my head this is the second time you sent a message i was thinking of sending but decided not to

ais523

for the polyglot challenge, we used to have language rundowns

Unrelated String

~~isn't it obvious how apt of a name it is~~

5

ais523

@Ginger a tautology would be "is X named X?"; "why is X named X?" is often a legitimate question

pxeger

but does it have super cow powers?

mousetail

13:36

The first rule of the tautology club is the first rule of the tautology club

ais523

the language rundown would be something like "this program prints 12 in language A, 11 in language B, 10 in language C", and so on

thejonymyster

idk if i didnt have the context of the song i might think A Pear Tree was some apl sort of thing

Ginger

@ais523 fair point

thejonymyster

@UnrelatedString took me a second

ais523

and the idea was to be able to finish it with "and a partridge in A Pear Tree"

Unrelated String

13:37

ahahah

pxeger

I'm gonna make a language called "gbirds" which prints "four call" for every program

thejonymyster

novelty languages r good

Ginger

laughs in CATHY

thejonymyster

@pxeger ok well thats a bit much lol i liked it when it was just the error

ais523

(which nearly any program will print, unless it's exceptionally long – the probability of a random program producing that error is over 50% well into the tens-of-kilobytes range)

thejonymyster

13:39

oh my god wait i just had a challenge idea but its probably not good so nevermind i didnt

mousetail

I hate that feeling

thejonymyster

i will talk about it though because maybe im wrong and you guys have ideas

it might be dupey though too

ais523

anyway, I put more effort into my one-off jokes than many people do, so I tried to work out, in a few minutes, what sort of interesting language would error out on almost any source code, which is where the checksum idea came from

pxeger

using CRC-32 for that checksum is pretty smart though

ais523

yes, needs to be something non-crypto-secure, otherwise brute-forcing it would take too long

and CRC32 was the first insecure hash function that came to mind

thejonymyster

13:41

every now and then ~~we all~~ i do the thing where i type a word like "apple" but end up typing "a[[;e" because my hand was too far to the right. i wonder if thered be a way to do a challenge where you like... correct those errors :P

ais523

I make similar errors but the offsets are more complex than just being one key to the right

thejonymyster

right

ais523

because I move my hands on the keyboard a lot as I type

thejonymyster

but for challenge i keep shrimple

it runs into issues if you try to take it mostly generally since like, actually detecting the errors in a string would be impossible

even with a dictionary of what is and isnt a word

pxeger

can we do a challenge that removes # from the end of the string, because I seem to press it accidentally every time I press the enter key

Unrelated String

13:44

ais523

you can just move the half of the keyboard that got destroyed by the bullets

4

thejonymyster

i guess itd have to be unconditional then?

ais523

(reference for those who don't remember the challenge)

thejonymyster

like "i absoutely 100% typed this with my hand in the wrong spot, fix the word plz"

and then guaruntee input wont have /?'"\|=+

(=+, /? and '" have keys to the right but i feel like that makes it a chameleon challenge along the lines of the cPS LOCK challenge)

ais523

the basic problem here is that you have a kolmo + a Jelly builtin

pxeger

13:47

The problem is it just becomes a string translation challenge

ais523

and I'm not convinced it's an interesting kolmo

pxeger

yeah basically ^^^

ais523

(and it isn't just Jelly that has the builtin in question)

thejonymyster

:P ah well there you go

ais523

hmm, we should have more challenges that are a trivial task combined with a ridiculously onerous I/O requirement – it's bad to combine restrictive I/O with an otherwise interesting challenge, but sometimes the I/O can be the challenge

thejonymyster

13:48

dont think its saveable then, thanks :-)

^^ interesting idea i like it

reminds me of this one idea that keeps bouncing around my head of "your output must be a valid input"

(which wouldnt be hard in most languages)

ais523

I've made at least one in the past, but it's likely an underexplored genre

also, esolangs suddenly start doing very badly when you make them do I/O beyond stdin/stdout/args

assuming that you aren't telling people to replicate the functionality of a standard program (which they will normally do simply by running that program)

thejonymyster

would "write a function, which when input into itself, does x" be self validating :P

ais523

OK, so challenge idea: "create a file whose size is provided by user input, but whose allocated size is < 1 MiB"

(i.e. a Lenguage assembler)

as far as I know, there isn't a standard program to do this (which is why Lenguage programming is such a pain) – maybe there is one and I've missed it

so in order to make it work you have to use fairly low-level APIs for interacting with the filesystem

it's even possible that C might win, which would be very surprising

mousetail

That term seems impossible to google

ais523

@thejonymyster self-validating is only appropriate if x is "return true", and then the task isn't interesting without an additional restriction

thejonymyster

13:58

thats what i suspected yeah

ais523

although you have got me thinking about what an esolang that has huge difficulty in returning true might look like

thejonymyster

>:)

ais523

maybe requiring specific user input

thejonymyster

ive just been learning a lot abt functional stuff lately :P like functions which return functions which take functions as input blabla

crazy world that programming is..........

ais523

I'm not sure what sort of language is best for messing around with that

pxeger

14:00

with functional stuff? or with a difficult esolang?

ais523

languages like Haskell do a huge amount of it, but are strongly typed which isn't much fun for the most serious messing around

functional stuff

Adám

Anyone up for some tacit JS?

avg=ㅤ=>ㅤ.reduce((ㅤㅤ,ㅤㅤㅤ)=>ㅤㅤ+ㅤㅤㅤ)/ㅤ.length;avg([3,1,4,1,5])
2.8

6

pxeger

lol

thejonymyster

next youll be telling me an array can have an array as an index oh wait

ais523

I'm trying to figure out the chaining rules

oh, this is monadic, so it's easy

pxeger

14:02

@thejonymyster tio.run/##y0osSyxOLsosKNEts/j/…

Adám

==> functions are the new => functions.

ais523

the only tricky part is interpreting + as quoted when you run the reduce, and that's presumably what (,)=> does in this notation

pxeger

@thejonymyster it just stringifies the array

thejonymyster

no i know i was just bein silly :P

pxeger

@ais523 (,)=> for dyads, => for monads, and ()=> for nilads?

Adám

14:03

Isn't there a way to build a 2-argument function using two =>s instead of (a,b)=>?

ais523

a=>b=>

Adám

But reduce doesn't like that.

pxeger

a=>b=> but that's curried, which is not what reduce wants

ais523

but the call syntax becomes f(a)(b) rather than f(a,b)

thejonymyster

@pxeger also this link is just my link again fyi

pxeger

14:04

dammit I'm so used to ATO auto-updating the link...

thejonymyster

right i figured :P

careful not to lose an answer that way

ais523

functional programming doesn't work well with tacit syntax because it's hard to disambiguate which operators are being called and which operators are being used as function arguments

Adám

@ais523 Uh, works fine in APL, and excellently in BQN.

ais523

Haskell's solution of quoting operators using parentheses is pretty clever, but wouldn't be great for golfing

pxeger

@Adám it's not truly functional programming though

Adám

14:05

BQN is truly functional.

ais523

I keep thinking that that should be BQM (so that it's APL+1)

pxeger

unless I'm mistaken, it requires extra syntax to use functions as "first class"

which, to me, makes them second class

Adám

@ais523 Indeed, but Marshall wants to reorder the alphabet so N comes before M, N having 3 strokes vs M's 4.

pxeger

Coincidentally (or perhaps not), there's the image format QOI which is a replacement for PNG, which is also +[1, 1, 2]

ais523

@Adám the alphabet did change order at some points in the past

pxeger

14:07

CMC: given a three letter string, increment the first two letters and add two to the last letter

(all inputs will be composed of a..x)

Adám

@ais523 Hebrew (from which the Latin alphabet ultimately derives) has M before N.

thejonymyster

@ais523 preposterous. The alphabet as is was handed down to us by the universe exactly as is on 1/1/1 (earliest date)

ais523

I thought the Latin alphabet was originally based on Etruscan, but with lots of influence from other languages

thejonymyster

or wait when does that one time standard start

ais523

let's look this up, I'm interested now

pxeger

14:08

@Adám Wouldn't it be more correct to say Hebrew and Latin share a common ancestor?

mousetail

@pxeger Python (a,b,c)=m;' '.join(*(chr(ord(a-1)),chr(ord(b-1)),chr(ord(c+2)))

Unrelated String

@pxeger not familiar with bqn but is the extra syntax required for handling functions as values, or just producing function values from functions-as-you'd-write-them-for-direct-use

pxeger

@ais523 looks like it was approximately Hebrew <- Imperial Aramaic <- Phoenician -> Ionic -> Latin

ais523

that seems reasonable

pxeger

@UnrelatedString yeah that's what I meant

Unrelated String

14:10

which one

ais523

I guess you could go down the Prolog syntax route, where basically everything is quoted by default

pxeger

oh I didn't read your whole message

the first one? idk

@ais523 Mathematica?

ais523

if you write A=1+2 in Prolog, it assigns the literal expression 1+2 to A because you didn't explicitly tell it to evaluate it

something which probably catches out people who are new to the language all the time

Adám

@pxeger BQN, 7: 1‿1‿2⊸+ Try!

ais523

@Adám what's the non-ASCII operator there?

pxeger

14:13

@pxeger Python, 42 bytes: lambda s:bytes(map(int.__add__,s,[1,1,2]))

@ais523 ‿ is for forming literal arrays (like , in most languages)

m90

@pxeger x86-64 machine code, 7 bytes: f: add DWORD PTR [rdi], 0x020101; ret (takes address in RDI and modifies the string in place)

Try it online!

ais523

oh, I thought that was an underscore

was interested in the other non-ASCII operator

Adám

Yes, and ⊸ curries the argument to +

ais523

ah, right, makes sense

pxeger

is it the equivalent of a tack in APL?

ais523

14:14

I wasn't expecting an explicit curry operator, but maybe that's necessary in a tacit functional language

Adám

@pxeger The alphabet used in modern Hebrew is indeed an offshoot of Aramaic. I meant the ancient Hebrew which is the same as Phoenician.

ais523

(unless you count concatenative languages like Underload, which have an entirely different set of chaining rules to APL despite being tacit)

Adám

@pxeger No, equivalent to ∘ in APL.

BQN has one for curry left argument and one for curry right argument (the little bar points to the argument). APL derives the meaning from the position of the array vs the function.

ais523

Underload always curries, but it has a function to reverse the direction of the curry

Adám

@UnrelatedString BQN doesn't really require any syntax to use functions as first-class values. E.g. you can create the array 1‿+‿2 just as easily as you can create 1‿1‿2 and if F is a function, you can pass it as argument to G with G f

Unrelated String

14:26

so yeah that would be unambiguously first-class then

Adám

In APL, you have to jump through hoops to do that sort of thing, though it can be done.

ais523

I think the "standard" golfing language solution to this sort of problem is to quote functions as string literals, and use string eval to do the unquoting

PyGamer0

15:24

@UnrelatedString @Zionmyceliaadamancy i dont know what i did here but i think these rules (for flax) might make sense: gist.github.com/PyGamer0/f313d41a4cafd07018ee646be33fc323

yeah i got rid of the 2,0 rules

and unparseable nilads are prepended to the accumulator

Unrelated String

i feel like in that case you would want to keep 2,0

PyGamer0

ok

@UnrelatedString done

Neil

16:21

@ais523 have you tried dd if=/dev/null of=$2 seek=$1 count=0?

Wheat Wizard

16:59

@ais523 It is, the Etruscans got it from the Greeks but they left their marks on it. For example you might notice that C, derived from Gamma, makes the same sound in Latin as K does in Greek. Which is because Etruscans didn't have a distinction between /k/ and /g/. So Romans used C for both until eventually inventing G for the /g/ sound.

mathcat

huh, I always thought the Romans got inspiration directly from the greeks

DLosc

17:40

See: c

mathcat

c: c

would've been ironic

DLosc

@pxeger Acc!!, 29 bytes: Try it online!

thejonymyster

oh yeah lol this is super easy in tiny imperative langs :P like bf is just ,+.,+.,++.

Steffan

18:17

@pxeger Vyxal, 7 bytes

Or if I/O can be list of codepoints, 4 bytes: ⁺»f+

Seggan

18:56

Woo 3K rep

Transcript for