The Nineteenth Byte

5:00 AM

@feersum Can you give an example of that besides speed?

Speed and memory are the main cases.

If you code in Java for example, there's no way to stop memory bloat, unless you stored all of the data in your program in a gigantic int array.

Thomas Kwa

@El'endiaStarman I think 1.5 bytes might be a good balance

feersum

I mean, there's no way to avoid the presence of pointers for every object, when they may be unnecessary.

El'endia Starman

I certainly won't be coding in Java. Probably Python, maybe C#. I'm expecting this language to be quite complex, and I have relatively good ideas of what it would take to do certain features in Python, but C# is kinda a big unknown. On the other hand, it would be a fantastic way to learn C#, I think.

Thomas Kwa

@Dennis No Bubblegum? :P

El'endia Starman

5:06 AM

@ThomasKwa So, maybe something like 1xxx denotes math, trig, graphics, etc? And 1000 would be something like switching instruction layers, as in (assuming math is 1001), 10001001 would be even more advanced/rare math operations.

quintopia

@El'endiaStarman I think one thing that makes a language powerful to think in is the ability to just try stuff. It's pretty much required that it have a powerful REPL, especially one that lets you write stuff that takes multiple lines, leave it there, execute earlier stuff in random order. Like editing LISP in emacs.

Thomas Kwa

@El'endiaStarman I'm fascinated by the idea that succinctness is just as important to readability as having each line be readable

@El'endiaStarman That sounds good!

El'endia Starman

@quintopia My plan for the parser should make it pretty easy to do a REPL.

quintopia

@ThomasKwa succinctness as measured in the number of distinct elements, to be sure

El'endia Starman

Yeah, one goal is to make the number of distinct elements small.

Thomas Kwa

5:07 AM

I think control-flow statements are inherently harder to read than data-flow ones, though

I think that's a reason that APL, despite having lots of elements, is acceptably readable.

El'endia Starman

@ThomasKwa Mind elaborating on the difference?

isaacg

@ThomasKwa Under this theory, where does Pyth land?

feersum

The problem with APL is that no two people speak the same dialect.

Thomas Kwa

@El'endiaStarman Hmm

quintopia

@isaacg Pyth is inherently unreadable just on the basis of not-very-memorable tokens

@isaacg hence why explanations generally separate chunks of it onto separate lines and explain what each does (the lack of inherent chunking also makes it unreadable)

isaacg

5:16 AM

Lack of automatic chunking is an issue

But many of the tokens are quite memorable

quintopia

of course

to you

isaacg

f is filter

m is map

Just not all of them

quintopia

another threat to readability is just knowing what variable assignments are being done in the background

like

isaacg

Yeah, that's getting changed in Pyth5

quintopia

you have to remember when you see "m" to say "that assigns members of the sequence to d"

isaacg

5:18 AM

to be easier to remember

It'll be assign to a in the first lambda, b in the second lambda, etc.

Thanks for the advice.

New Main Posts

0

Q: Find a one liner in vb for case function when case increments every 5 natural numbers

Having a number greater or equal to 10 and having a max case when is equal to 70 like: Select Case number Case 10 To 14 Return 10 Case 15 To 19 Return 14 Case 20 To 24 Return 17 Case 25 To 29 Return 20 Case 30 To 34 Return 25 Case 3...

quintopia

...so now you have to count how many lambdas in you are?

isaacg

What nesting depth you're at, actually.

quintopia

that's what i meant

you know what would help more than any language changes?

a visual editor, with all the tokens given longer names

El'endia Starman

@quintopia, have you taken a look at Minkolang's online interpreter? If so, what did/do you think?

quintopia

5:21 AM

you type "m" and "map(a:" appears or something like that

@El'endiaStarman no

Mitch Schwartz

this is amusing - gibberish passed off as a brainfuck answer

3

A: When is Thanksgiving?

Brainfuck, 84 Bytes This one's significantly shorter than my Jython version. Computes the date based of of an offset from the beginning of the year and adds one for leap years. ,+++[>++++++++++<-]>[>+++++++++++<-]>>++++<<[->+>-[>+>>]>[+[-<+>]>+>>]<<<<<<]>>>>[<<<<+] Edit: 97 Bytes Forgot to s...

El'endia Starman

If you don't mind... --> n3;1-1Md1%-2;N. outputs the largest square less than N^3, where N is given through input.

Mitch Schwartz

i love how he continues to defend it after being called out

isaacg

@quintopia I'll look into.

El'endia Starman

@ThomasKwa /me waits patiently

xnor

5:37 AM

does haskell have a way to count elements of a list that's shorter than length$filter(==x)?

or sum[1|y<-l,y==x]

quintopia

@MitchSchwartz apparently deleted?

isaacg

@quintopia Apparently, a 22k rep user telling them their code obviously didn't work and flagging it as not an answer helped.

Thomas Kwa

@El'endiaStarman After some thought, I'm not so sure about it

Mitch Schwartz

@quintopia ^

Zach Gates

Can you link that answer?

nevermind

I scrolled up

facepalm

El'endia Starman

5:48 AM

My current ideas for the 1xxx half-byte instructions (libraries): meta, math, string, arrays (including lists), graphics/images/audio, internet/network, physics?, ???.

"meta" meaning that "meta"-"math" will give access to even more math functions (another 128 more).

"meta"-"string" could be something else, though.

quintopia

meta string could be something like python's "string", adding functions and constants to the standard "str"

El'endia Starman

More likely those would just be in "string".

Each of these libraries has 128 available commands.

File operations and I/O will almost certainly be exclusively in the ordinary (128) instruction set.

Thomas Kwa

Have you thought about making the binary for commands in, say, the "arrays" block shorter after another "arrays" command?

El'endia Starman

I'm not sure how well that would work in practice. Do you mean something like \arrayCommands(\commC(\commA(x\commB(y|, where | signifies the end of array commands?

On further thought, if I want to leave the entire alphanumeric set for identifiers, then that restricts the number of ordinary commands to about 30 or so. Maybe more, depending on what I do with symbols like &.

Thomas Kwa

6:07 AM

@El'endiaStarman I don't know exactly how

Just an idea

quintopia

@El'endiaStarman time to reconsider my proposal to automatically compile identifiers to non-alphanumeric bytes :D

El'endia Starman

I'm not sure exactly what improvement you expect from that.

quintopia

@El'endiaStarman Since you're using tokens anyway, why are you even worried about ASCII considerations?

Thomas Kwa

@El'endiaStarman You're throwing away 52 of 256 characters.

quintopia

carrot.jpg

El'endia Starman

6:12 AM

@ThomasKwa 52/128, actually, because of the 1xxx libraries.

quintopia

62 even

Thomas Kwa

Let's call it 56 out of 256

log(256)/log(256-62) is 1.053

El'endia Starman

But consider this: \nLargest(L3 only works because identifiers are alphabetic.

quintopia

does it? i thought \nLargest got mapped to a single byte?

Thomas Kwa

That means your code will be 5% longer without using the alphanumerics in the regular command space.

El'endia Starman

6:15 AM

Otherwise I'd have to do something like \nLargest("L"3, 5 bytes at best, or something like \nLargest(3\retrieve(L, 4 bytes at best.

quintopia

yes, perhaps. but in bytes, \nLargest(var_L3 may not even be larger than \nLargest(L3

Thomas Kwa

^

El'endia Starman

\var(L would be two bytes.

quintopia

yes, but \var_L would be 1!

El'endia Starman

Not if I want arbitrary identifiers.

Thomas Kwa

6:17 AM

Why would you want that?

quintopia

hurray for automatic compilation of arbitrary identifiers to bytes!

El'endia Starman

@ThomasKwa I've said before that I want to actually use this for projects.

Try writing, say, Tetris in this language.

Thomas Kwa

You could have both

quintopia

carrotetris

Thomas Kwa

Half a dozen reserved variables, and arbitrary other ones

quintopia

6:19 AM

or not even that!

again, the magic of compilers is: you don't have to decide what the reserved bytes are called in advance. you can name them when you use them!

in fact, if you're smart, you could make a compiler automatically reuse bytes with different names when it knows the other name won't be used again for instance when it goes out of scope

El'endia Starman

I'll be constructing an AST, so I can definitely do that.

Okay.

How should I distinguish var_L, 3 and var_L3?

Assuming that L3 is an identifier I used.

Thomas Kwa

I think I might win Thanksgiving with TI-BASIC. 14 bytes

El'endia Starman

6:36 AM

Hmm. That might be something like \holiday("Thanksgiving\getInt(\input. 2 or 2.5 bytes for \holiday, 13 bytes for "Thanksgiving, and 2 bytes for \getInt(\input. 15ish bytes altogether. Doing it the math way would likely be shorter.

Now, if I implemented \holiday( such that it can do matching on partial names, I could, perhaps, get away with \holiday("Than\getInt(\input. 7ish bytes.

quintopia

6:48 AM

@El'endiaStarman separate them with a comma. you could even make the comma mandatory. it's only for the parser anyway.

El'endia Starman

hmm

In the compiled byte code, how would I separate them?

I've been planning to use the hexdump as the guide for how many bytes a program is.

phase

@El'endiaStarman a hex dump would give you how many bytes it is, since that's exactly what a hex dump is

El'endia Starman

@phase Aye, but I mean in the sense that \map( would not count as five bytes, but one.

phase

@El'endiaStarman You could just use a single byte char for that

feersum

Why do all commands begin with `\`? Is it TeX-based?

El'endia Starman

7:00 AM

I plan to have an interpreter/compiler that turns \map( into, say, 0x56.

@feersum I'm drawing inspiration from TeX, yes.

feersum

Is teh interpreter in it?

El'endia Starman

`\` is like, the universal escape character. Seems like a good choice for saying "what follows is a command".

Noooo, I wouldn't use TeX for programming. I only know LaTeX.

2 hours ago, by El'endia Starman

I certainly won't be coding in Java. Probably Python, maybe C#. I'm expecting this language to be quite complex, and I have relatively good ideas of what it would take to do certain features in Python, but C# is kinda a big unknown. On the other hand, it would be a fantastic way to learn C#, I think.

phase

j.[$[$,]{n_0}%]\[$,r]\(;${n}d] negative input to positive input range, only 30 bytes. ;-;

@El'endiaStarman Why is Java worse than C#?

El'endia Starman

@phase I barely know any Java.

phase

> it would be a fantastic way to learn C#

sounds like you don't know C# either

El'endia Starman

7:03 AM

Companies in my area want C# programmers and I've heard often that C# is a good language.

phase

It only runs on Windows, so not every can use your inter/compiler

TanMath

when did mego last come?

phase

@TanMath Wasn't on all day today (I think)

He pushed to Seriously 14 hours ago

El'endia Starman

@phase I'd make an online interface for it, like I did with Minkolang.

(When I say online interpreter, I really mean online interface.)

phase

@El'endiaStarman But what about things that won't work online? (Such as infinite loops)

TanMath

7:07 AM

@phase that is sad.. I want to know something about Seriously so I can finish up two of my seriously answers!

El'endia Starman

@phase Already figured that out. :D

phase

@TanMath I may be able to halp

TanMath

@phase how can you do i[1:] in seriously? i is a list...

phase

@TanMath what about reversing the array and popping a value off?

TanMath

@phase code pls? what if there is already stuff on the stack?

El'endia Starman

7:13 AM

Reverse, pop, reverse.

phase

I can't find an operator to pop an array, but R is reverse

quintopia

Why is there no wiki page for Seriously :c

phase

@quintopia github.com/Mego/Seriously/blob/master/commands.txt

@quintopia seriously.readthedocs.org

quintopia

so?

phase

What kind of wiki page were you looking for?

El'endia Starman

7:18 AM

X pops, I think.

And I found this by looking through the code. I must be a madman. :P

quintopia

no

@TanMath d pops an array, dequeues it, pushes back array and dequeued value

so

RdXR

TanMath

@quintopia gracias!

El'endia Starman

OH. DUH. Seriously has arrays!

I was thinking of popping off the front of the stack...

phase

@El'endiaStarman claps

quintopia

the seriously github appears to have no licensing info

phase

7:23 AM

@quintopia (c) Mego 2x15

quintopia

in what file?

phase

@quintopia no file, just default copyright on github

quintopia

eh it doesn't matter. if it's not explicitly free-licensed, it's copyright

oh well

I was going to make a wiki page for it

I still could, but it would just be all links

(a page with the command list would be better)

@Mego!

Speak of the devil

Anonymous

@TanMath [1,2,3]p -> [2,3],1

Anonymous

Actually that looks broken right now

Anonymous

7:30 AM

So you'll have to do [1,2,3]RdXR

quintopia

@Mego figured you would have a shorter way. it'd be easier to read the commands file if the different type signatures were on separate lines. or something.

phase

and hex values too

Anonymous

That's on the list

quintopia

Idea:pure functional language where every function has arity 2, and the source is just a compact representation of a binary tree.

Anonymous

Idea: stop having ideas :P

El'endia Starman

7:43 AM

Aaand there's mortarboard #2 on SO.

quintopia

what is mortarboard

El'endia Starman

Hitting the rep cap of 200 from upvotes.

Accepts and bounties not included.

Anonymous

Doesn't have to be from upvotes

Anonymous

You just have to hit 200

El'endia Starman

100 edits sure is a lot... :P

But yes, not strictly upvotes.

Anonymous

7:47 AM

I got my first mortarboard largely due to bounties

feersum

Is there a negative reputation cap?

El'endia Starman

Not that I know of.

Anonymous

Idk ask Geobits

El'endia Starman

Any serial downvoting would be caught and corrected, though.

feersum

We need to find out.

phase

7:48 AM

I'll start posting questions

feersum

What could an answer do to get 100 downvotes?

Without getting deleted.

Anonymous

Abuse a loophole

feersum

Might get lots of upvotes anway.

Anonymous

Specifically abuse a loophole through rules-lawyering, one that later gets closed in the standard loopholes thread

feersum

We could have a mod stand by and continually undelet it.

If the question says "print the square of x" and I print the text "the square of x", that could be annoying enough.

phase

7:50 AM

-3

Q: My Squiggly Lamp

While moving, I broke my lamp. Now, I need a new one. It's your job to make me a lamp! I'm not sure what size I want, though I know I want a squiggly one. Your program/function must take in a number input, and print out a lamp with that many squiggles/bumps. Examples: Input:2 Output: /--\ ...

code-golf ascii-art

has 8 downvotes ;-;

Anonymous

@feersum That's a standard loophole

phase

feersum

@Mego It needs to be an obvious and annoying loophole to get downvotes.

Doesn't matter if it's in the meta post.

One that's not in there is probably too weird and will get upvotes.

Anonymous

@feersum If it is a standard loophole, it'll get deleted

feersum

This is why we need @AlexA. to sit at the computer and repeatedly click undelete.

Sp3000

7:56 AM

You need to add "obviously not well golfed" in there somewhere, for extra irony (loophole abuse but can't loophole abuse properly)

xnor

Mini-challenge: Write the simplest-sounding task that takes more than 50 chars to golf in Python.

El'endia Starman

Find the first place that Li(x) < pi(x) (I think that's the right direction).

Nevermind that it hasn't been found yet...[whistles]

Sp3000

Not at a computer so can't test: Sort a list of newline separated numbers via STDIN, outputting newline separated to STDOUT

Anonymous

@xnor Print out all the ASCII alphanumeric characters in ascending order, sorted by ASCII ordinal

quintopia

xnor: write a program that prints "AzpGoengJuifiPelfnmAAAjfoeow[qasdghyiviuYVGIYGUVkjdheygqie;GFUIADEFeiugf;qe"

feersum

8:05 AM

@Sp3000 That should be doable, in 3 at least.

print(*sorted(map(int,open(0))),sep='\n')

El'endia Starman

@Mego import string\nstring.digits+string.ascii_letters (Where \n is a newline, of course.)

Sp3000

What's open(0)?

feersum

That opens stdin in Python 3.

Anonymous

@El'endiaStarman string.ascii_letters is lowercase first

quintopia

@feersum open(0) o.0

oh

xnor

8:06 AM

@feersum it reads multiline STDIN?

El'endia Starman

...dang

Anonymous

@Sp3000 print'\n'.join(sorted(map(int,open(0))), 39 in 2

Sp3000

Darn, didn't know about that one. Was hoping for something that would require try/except or os for multiline input

feersum

Using a file as in iterable always splits it by lines.

@Mego open(0) doesn't work in 2.

Sp3000

@Mego Not sure you can join ints by str there

xnor

8:08 AM

@feersum i get OSError: [Errno 9] Bad file descriptor

is it for a specific filesystem?

@Sp3000 i was also thinking along the lines of taking multiline input

Anonymous

@feersum Darn

feersum

@xnor That's odd. fd 0 being stdin is a Linux thing, but they hacked it to work in Windows too.

Anonymous

@phase Your constant wish has been granted: commands.txt now has hex instead of decimals for ordinals

feersum

What OS?

xnor

windows 7

phase

8:09 AM

@Mego yasssssssssss

feersum

Works for me.

phase

@xnor I got that, try Command Prompt instead of MinGW or whatever you're using

xnor

i'm doing it from IDLE

phase

well..... don't do it from idle....?

feersum

^

Anonymous

8:11 AM

^

phase

^

xnor

4: well..... don't do it from idle....?

feersum

@phase agreed with himself so it's invalid input.

phase

^

Anonymous

Hey @phase, I'm trying out a new feature in the IDE, can you help me test it?

phase

8:13 AM

@Mego Chrome shows you what links go to when you hover over them

Anonymous

@phase Whatevs

Sp3000

@xnor Tell me if you get it working, I'm curious :)

Anonymous

Now, sleep time

Anonymous

Early morning cooking tomorrow for turkeyday

phase

gobble gobble

Zach Gates

8:15 AM

^

Anonymous

I'll try to make Seriously suck less after the food coma ends

Anonymous

But in all actuality it'll just suck more

Anonymous

\o/

phase

Suckerously

Sp3000

Ah Thanksgiving, one of two days where... nothing much changes over here in comparison, really (along with Halloween)

Alternative for the 50 byte challenge (also not sure if it'll work): Given a float via stdin, output to STDOUT rounded to 3 sig fig

o_O okay...

feersum

8:21 AM

@Sp3000 print'%.2e'%input() ? Or am I missing something?

xnor

one second, i think i messed up

@Sp3000 never mind, it works in cmd

how does I terminate the input though?

feersum

ctrl-Z (in windows)

Sp3000

Should probably have specified not scientific notation

xnor

@feersum thanks

the open(0) thing is a useful trick i didn't know about. you should write a tip

Sp3000

^

feersum

8:24 AM

I'm not very proficient in Python 3. I don't feel qualified to write tips about it.

quintopia

oh okay. well...do it anyway?

feersum

1 and 2 are the standard Unix file descriptors for stdout and stderr, FYI.

Sp3000

I await the day we see print(...,file=open(2))

quintopia

for printing errors?

Sp3000

And not halting, unless there's better

xnor

8:30 AM

how about "Determine if one list is a contiguous sub-list of another list"?

quintopia

should be easy?

something like

feersum

`a[1:-1]` in `b` ? :P

quintopia

yeah that

Sp3000

@xnor Might be doable with string reps in Py 2, but can't try atm

xnor

ah right

feersum

8:32 AM

Depends on what's in the lists.

quintopia

@feersum you fast

Sp3000

nvm phone lag beaten by feersum :P

xnor

it's weird how many more built-ins there are for strings than lists

Sp3000

You probably need to tack a few spaces so you don't get false positives though, eg [2] in [12]

feersum

I think it can be done for real in under 50 anyway.

f=lambda a,b:b and(b[:len(a)]==a or f(a,b[1:]))

Sp3000

8:35 AM

@xnor One thing that gets me all the time is that I forget lists don't have find :/

xnor

@feersum that gives the empty list on False, I was assuming actual bools, not truthy/falsey

are there enough chars to patch that?

also, the empty list is a sublist of the empty list

We can reshuffle it.

What's in the lists?

let's say integers

f=lambda a,b:b[:len(a)]==a or b>[]and f(a,b[1:]) seems to work.

Sp3000

8:40 AM

Does something like a+['']>b>=a work?

xnor

that's nice

I forgot that startswith replacement

Sp3000

What's larger, strings or functions in Py 2?

xnor

strings

Sp3000

Darn, was hoping to handle strings as well if id worked :/

How about dicts/tuples/lists?

xnor

[0, <built-in function id>, {}, set(), 'a']

Sp3000

8:46 AM

:/ does anything beat strings? (Apart from I think Ellipsis)

feersum

Where do lambda functions or generators fit in?

xnor

[0, <built-in function id>, {}, <function <lambda> at 0xa72da0>, <listiterator object at 0xc65000>, set([]), '']

feersum

You can also compare to random objects like files...

xnor

[None, 0, <built-in function id>, bytearray(b''), {}, <function <lambda> at 0xc64440>, <listiterator object at 0xa732c0>, set([]), '', xrange(3)]

Sp3000

Home now, apparently I was wrong about Ellipsis (and NotImplemented doesn't help either...)

I thought there was something bigger than everything though...

(just like how None is smaller than everything)

xnor

8:53 AM

[None, 0, NotImplemented, <built-in function id>, <built-in method viewkeys of dict object at 0xbabf40>, bytearray(b''), {}, frozenset([]), <function <lambda> at 0xbab760>, [], <listiterator object at 0xbaba60>, <memory at 0xbabc60>, set([]), '', xrange(3)]

El'endia Starman

float('inf')?

Sp3000

That's a number, so it's smaller than str actually

xnor

[None, 0, inf, NotImplemented, <built-in function id>, <built-in method viewkeys of dict object at 0xba99c0>, bytearray(b''), {}, frozenset([]), <function <lambda> at 0xc62920>, [], <listiterator object at 0xa71140>, <memory at 0xbab180>, set([]), '', xrange(3)]

feersum

Should probably read the code at this point :P

Sp3000

On it :P

feersum

8:55 AM

Could be getting an order based on random memory addresses.

lol reference counting for True and False?

Sp3000

Hmm I'm getting slightly different results from xnor for a few in the middle, so you might be right...

Oh nvm, xnor's got the viewkeys function, not an actual dict_keys([])

xnor

the docs say "objects are ordered consistently but arbitrarily "

Transcript for