The Nineteenth Byte

6:06 AM

yeah, that works

now THAT is copy on write

as opposed to copy and write

the crux is that no copy happens until a write occurs

well, that's not strictly true

we also have a condition that a copy won't occur if its also the only copy

@NathanMerrill if it's the only copy then there was never a copy :)

basically it only copies when needed

its basically "copy on 'clone and write'"

but that's a bit of a mouthful

copy on write is pretty much always more efficient than copy on assignment, right?

@NathanMerrill no, because there is overhead

Shared is another indirection layer

before it was just

self.l, now it's self.shared.val

6:11 AM

but in terms of number of copies, copy on assignment will always produce equal or more copies

yes

but there's more things in the world than just copying

access time became slower :)

true

you can also have different strategies

for example if you very often copy some buffer and change only a couple elements

you can have a "copy on nth write"

and until the nth write it stores the changes into a dict

ok. So, I think I'm going to use the word "immutable" to indicate that all modifying functions are copy-on-write

unless there's a better word?

again, they're not copy on write

your strings and integers are immutable

they are never modified

there are no modifying functions

6:14 AM

right, but what other word indicates "you can't change this object, only create a new copy with a modified value"

that's just immutable

@NathanMerrill if I write 3 + 5

even though a copy-on-write list could be modified

in the case that there is only 1 copy

a list is not immutable

lists can be modified

a = [1, 3, 5]; a[1] = 42

a COWList :)

if I write 3 + 5 (or plus(3, 5)), it doesn't modify 3 or 5

it just returns a new (also immutable) value that is the sum of the two

6:16 AM

but is there a better word than immutable?

no

this is exactly immutable

(for copy on write)

I don't think that anything you're doing pertains copy on write

while I see the difference, I don't think there's a word that is closer

and you should probably stop using the term

immutable is a value that doesn't change after initialization

6:17 AM

no, copy on write is definitely what I want.

want or have?

want

that's what I mean

but copy on write is only relevant for mutable objects

things that can be written AKA changed

right, but if somebody makes a class

classes are mutable

6:19 AM

then I need some way for them to indicate that this a copy-on-write class

could you give an example of what a copy-on-write class would look like?

sure

because in order for you to do that you already need to have a model of copying

classes will have a clone function

Python has __copy__ and __deepcopy__

which are used with the copy module

copy.copy calls __copy__ on an user-defined class

and copy.deepcopy calls __deepcopy__

@NathanMerrill but that is the simple part

now you need a protocol that determines what is 'writing', and what is 'reading'

if I have some class A, with an object a = A()

obviously a.l = [] is writing

but is a.l.append(1)?

6:22 AM

oh, that's part of my typing system

how do you detect that?

functions are by default read only

you have to explicitly declare that they mutate

so, writing to a class variable can only be done in a mutating function, as well as calling a mutating function

ok, then I guess it can be done

or...

you have to consider writes/reads from outside of functions

e.g. in the main body

only from a "pure" functional standpoint

another function could modify a variable I was reading

What is a good website to learn code for an absolute beginner?

6:26 AM

so, my read function doesn't necessarily always return the same thing

oh, I understand. Yes, if somebody had a publicly visible class variable, writing to it would make the function a mutating

(I don't have static variables outside of constants)

though, I am considering removing that in favor of logging

@JesterTran learnpythonthehardway.org/book

learn Python

@orlp hmm, the hard way...

and why Python?

Python is amazing for beginners

I haven't read the book, but I do agree that Python is the best first language

how about C?

6:30 AM

it's readable, simple and doesn't have a lot of wtfs that other languages have

@JesterTran absolutely not

@JesterTran it is full of the "wtfs" that orlp mentions

how versatile and useful is python?

well, all languages can pretty much be used anywhere. That said, python is mostly used for web development and scripts

is there anything that a turing machine cannot do

scripts being small programs that process data for one particular purpose

6:32 AM

but a regular programming language can?

no

@KennyLau yes

if you're looking for Android development, then you could learn Java.

@orlp like?

turing completeness refers to computational capability

but a brainfuck program can never show a picture on the screen

6:33 AM

sure, a turing machine can't access the internet

because it doesn't have the I/O capabilities

well, a java program can't show a picture on the screen either

there is more to a programming language than computation

input -> computation -> output

ok, I will learn Python

thanks, everyone

good choice!

6:34 AM

I'm referring to computation

is there anything that turing machine cannot compute?

yes

@KennyLau yes

but it can compute anything that our computers can compute

but a regular program can?

Undecidable problem

In computability theory and computational complexity theory, an undecidable problem is a decision problem for which it is known to be impossible to construct a single algorithm that always leads to a correct yes-or-no answer. A decision problem is any arbitrary yes-or-no question on an infinite set of inputs. Because of this, it is traditional to define the decision problem equivalently as the set of inputs for which the problem returns yes. These inputs can be natural numbers, but also other values of some other kind, such as strings of a formal language. Using some encoding, such as a Göd...

6:34 AM

"but a regular program can"

@KennyLau that was added after I alt tabbed

either way, define "regular program"

what do you think of codeacademy's course for learning Python?

and define "compute"

@JesterTran I've heard bad things of codeacadamy

Well, we always stress on a language being "Turing-complete"

@JesterTran I had my sister take it. She thought it was a bit slow.

6:36 AM

as if Turing-completeness by default is already the highest level

but that said, its really not too bad of a resource

@orlp @NathanMerrill I see.

just realize that you are going to really learn to program when you come up with something you want to develop and actually make it

you really want a book

shop.oreilly.com/product/0636920028154.do

this is what I learned Python from

but I can understand not wanting to buy a book

so then you can use learnpythonthehardway that I linked, which are free

yeah, I'll probably learn it the hard way

6:38 AM

now, there is another option

and that is to learn scheme

through SICP

this is less about programming and making real-world programs, but is very fundamental in how compute programs work, and computation

@JesterTran Or you could just get someone to answer your questions

like "how to do <stuff> in python"

and then read through the Python documentation

@orlp I haven't used that, but is it really beginner friendly?

@NathanMerrill yes and no

it's beginner friendly for a freshman CS student

it requires some pre-existing mental fortitude, maturity and a lot of investment

sounds cool

@JesterTran if you have any question about Python you can ask them here

@JesterTran how old are you?

6:42 AM

ok

16

(don't answer if you are under 13)

@JesterTran I wouldn't recommend SICP then

chat mini-challenge: given two numbers on the first two memory position, put the maximum of them in the third memory position

@KennyLau in what language?

or machine

or context

oops

in brainfuck

6:47 AM

can the first two be destroyed?

You can use this template

yes, they can

@orlp what about unshared as a keyword for copy-on-write?

@NathanMerrill you mean shared?

no, because if they were shared, then they modifications in one would affect the other

I dno

I don't really care either =/ too long on the same subject :P

6:52 AM

ok

I'll sleep on it

thanks for the discussion anyways

@Sherlock9 try my challenge?

Sherlock9

Link/onebox?

chat mini-challenge: given two numbers on the first two memory position, put the maximum of them in the third memory position in brainfuck

Please use this template

7:08 AM

wait something bugged

Sherlock9

7:33 AM

Weirdass Internet connection

Hmm. I don't know BF very well. I'll see what I can do

@KennyLau do we need to handle 0?

No, you do not.

(I don't see how the code will be different)

A: Sandbox for Proposed Challenges

7:49 AM

fatiherikli.github.io/brainfuck-visualizer/…

not terribly elegant

but it works

New Sandboxed Posts

0

Brainfuck-golf: find the maximum of two numbers code-golf You will be provided with two numbers on the first two memory cells, and you will write a code in brainfuck to put the maximum of the two numbers on the third memory cell. You may use this template and this template to test your code. ...

nice. how does it work?

>[<<+>>->-[>]<<]

the [>] is crucial here

it offsets one position at some point when one of them hits 0

I see

and then?

then it adds whats left

it's a bit weird but it works

7:56 AM

lol I don't get it

8:18 AM

The part before [>] increments tape[i+2], decrements tape[i+1] and tape[i]

If one is zero, [>] runs fewer times or something, making the loop stop

Hello

8:47 AM

@KennyLau I have a slow but elegant algorithm

fatiherikli.github.io/brainfuck-visualizer/…

it can probably be optimized quite a bit, just made it work

just running it should already tell you how it works

what do people think about length restriction fastest code challenges?

are they allowed, to start with

hi @orlp

hi

@orlp I like that

for example, fastest sat solver in under 150 bytes

@Lembik hi part of all

8:50 AM

hi @zyabin101

any thoughts?

@HelkaHomba That has been in the works for quite a few years already

flawr

9:06 AM

@LuisMendo Now I did, great news!

9:25 AM

any views on the acceptability of length restricted fastest code competitions?

I can imagine someone will hate hard length restrictions

@Lembik Don't think it's a good idea

@MarsUltor thanks! Why is that?

I am trying to avoid the problem where longer and longer solutions in lower level languages always win

Longer/non-esolangs have a disadvantage vs golfier ones

@MarsUltor but golfier languages are slower typically right?

i.e. golfy languages, although slow, can use a more efficient algorithm because they can perform more functions

9:29 AM

@MarsUltor right.. but SAT solving doesn't have a very fast algorithm

so I thought the compromise could be interesting

If the length limit is too low, non-golfy languages can hardly implement anything

@MarsUltor true.. setting the limit correctly is importsant

and hard to guess a priori

I was hoping people might help me with that on chat :)

Most C#/Java golf solutions are around 300, I think

Python/Ruby/JS would be around 100

ok.. so how about this concrete proposal. Fastest code competition to write a SAT solver in <= 350 bytes?

python is dog slow after all :)

Pypy is fast though

9:31 AM

I use pypy a lot and the situation is complicated

@Lembik I guess

your code has to be pure python to start with which excludes most of the point of python for people doing scientific work

that is the vast set of libraries

@Lembik Yeah

and second, even in pure python it is something 5 times faster and sometimes a little slower

it's hard to predict

It works well mainly when a function is called many times, because it caches the results for certain inputs

9:33 AM

they could just implement numpy in pypy but it goes at 1 mm/year as they just don't have the coders

pypy also works very well for simple and straightforward implementations of classic algorithms when you ignore all of the built in helpful functions like itertools

monking

the devs say they hate itertools :)

@Lembik what time zone are you in?

This one :)

@EᴀsᴛᴇʀʟʏIʀᴋ monring

9:34 AM

you?

...

For me, it is 02:34.

12:35 here.

always or just right now?

...now.

for me it's always just before lunch :)

9:35 AM

@zyabin101 noon?

@Lembik ....

offset to UTC?

@EᴀsᴛᴇʀʟʏIʀᴋ UTC+3

@MarsUltor you are right of course that some languages will be better suited than others. That feels sort of inevitable though

what happened to GMT!!

it died in hell. Now answer me.

my mother told me not to give away personal data like which planet I was on :)

A: Things to avoid when writing challenges

...

So did mine.

Last night. So I sanitized exactly 1 occurrence of my full name. shrugs

hey @Mego

Anonymous

9:37 AM

@aditsu @Adnan codegolf.stackexchange.com/questions/80196/…

Anonymous

11

Requiring minimum scores In short, don't post a code golf that says "Your code has to be shorter than 100 bytes" (the same applies to any other winning criterion). The usual motivation is that the author found some code somewhere and challenges people to beat the score of that code. That's a fin...

Yeah, I think that one is a dupe.

Anonymous

It's very clearly a dupe

@Lembik are you going to answer me or should I go back to bed ~~or playing nethack~~?

Well not a dupe, but a subset. Which still counts for closing as a dupe here.

Anonymous

I cannot fathom why aditsu would use his gold-hammer to reopen it

aditsu

9:41 AM

@Mego I wholeheartedly disagree

gold-hammer? what is that?

Anonymous

You have a gold code-golf badge, which means that things like closing as dupe and reopening are immediate for you

Anonymous

You don't have close votes; you have a close hammer (same for reopen)

aditsu

oh I didn't know about that

afk for a while..

Anonymous

I wholeheartedly disagree with taking unilateral action to reopen a challenge without a discussion on meta first

lol

@Mego I think he doesn't know what he did. :P

Anonymous

9:43 AM

You've been around for 6 years and didn't know that having a gold tag badge meant you have a close/reopen-hammer?

Anonymous

(not doubting you; just a bit surprised)

I didn't know I had a christmas present waiting in the closet for at least 4 days, and it then was destroyed by the cat.

Anonymous

The AI Area51 proposal needs more love

> You have to download a total of 3,230 M. This download will take
about 2 minutes with your connection.

lol

9:48 AM

wishes that was true

@Mego Already committed, sorry.

@MarsUltor if it is around 7 where you are, wtf are you doing up so late every day?

> first commit

@EᴀsᴛᴇʀʟʏIʀᴋ ?

late?

You are sometimes active on what is 18:00 PST, but isn't that like 23:00 for you?

11pm is late?

> first commit

@EᴀsᴛᴇʀʟʏIʀᴋ Do you sleep at like 9pm every day?

9:50 AM

if that was me, then yeah.

@KennyLau about there

Anonymous

@EᴀsᴛᴇʀʟʏIʀᴋ I didn't notice you had committed because your network account still uses your old username :P

Why is Area 51 pretending to be git

@Mego lol

@MarsUltor First commit?

Oh wait

Anonymous

9:50 AM

@MarsUltor It's not. There are multiple meanings for "commit".

@Mego I know

@Mego fixed hopefully.

> "To help Marky Markov find a loving home..."

lol I forgot I wrote that

mınxomaτ

@HelkaHomba So now we can have software and hardware exploits on Android phones. Neat ;)

@EᴀsᴛᴇʀʟʏIʀᴋ Well, 1. because I can, and 2. in twelfth grade in Australia you probably usually stay up late doing homework/assignments anyway (not that I do)

Fatalize

@Mego the level of questions is too low to attract experts

9:53 AM

@MarsUltor okay

Anonymous

@HelkaHomba They've been "in the works" for many years (essentially since smartphones started getting really popular). I remember seeing a PopSci article about modular phones around 2010.

> twelfth grade

:O

Anonymous

@Fatalize What is that supposed to mean?

In Russia, there are eleven grades.

It should have that twelfth grade :D

@zyabin101 In New Zealand, there are thirteen (their first grade is preschool)

Anonymous

9:55 AM

hehehe @ the hovertext on this one

not actually?

:P

> Infinity/2
Seriously?

Fatalize

@Mego I'm not interested in answering questions like "What's the difference between AI and genetic algorithms?" which make no sense in the first place

> Infinity/2
Infinity

sorry smbc, you have failed.

@Mego Related :P

Anonymous

9:57 AM

Thought: is there any string (finite or infinite) that cannot be expressed finitely (as a result of a finite number of operations in some programming language)? In other words, is there any potentially-infinite string whose Kolmogrov complexity is infinite?

[3] and []

Seriously?

this is weirder