The h Bar

vzn

2:28 AM

via reddit, spking of software vs physics, the two are increasingly synergistic.

Physicists ≠ Software Developers / De Castro, AMVA4NewPhysics

DanielSank

5:03 AM

Hi, everybody.

@BernardMeurer Peachception.

Hey, @BernardMeurer what the hell's going on? Government turmoil and a mosquito-carried virus that causes birth defects. Scary.

Bernard Meurer

@DanielSank heya

Idk what's going on, I'm just getting out of here ASAP, don't want to risk getting stuck later on

DanielSank

What are people saying there about the olympics?

The talk online is generally that lots of people travelling to Brazil could lead to a world health disaster because of zika.

Bernard Meurer

Zika isn't the trouble people are making it look like

I only know one person who got it and:

1. I know a lot of people

2. I live in Rio, the actual Olympics city

We've been fighting the mosquito for a long time, mostly because of Dengue, and we can handle that

DanielSank

So the news coverage is overblown in your opinion?

Bernard Meurer

Here people almost forgot the Olympics are happening, our politics is better than House of Cards so we focus on that

@DanielSank Very much so, it's like H1N1; "IT'S THE END OF THE WORLD EVERYONE WILL DIE", two months later no one even knows what that is again

DanielSank

5:14 AM

Well, people worry more when the pathogen affects babies.

Oh, @BernardMeurer I want to tell you the best bug I've ever seen. It happened today.

Bernard Meurer

Yeah, I guess, I'd be far more worried about the athletes in rowing and things like that

@DanielSank Bug story time! Yay!

DanielSank

@BernardMeurer Oh, that's interesting. Will talk about that after bug story.

Ok, here it goes...

I was using the Google Cloud SQL to set up a database for some stuff at work.

There are several ways to connect to a database in the cloud. For example you can use a direct TCP socket connection from e.g. mysql client to the cloud database server.

However, to do this, you have to white-list your IP address, and there's a really annoying constraint that if the cloud SQL instance uses IPv4 or IPv6, the client has to make a connection from the same kind of IP address.

This was a problem, so I decided to use a proxy provided by Google.

The proxy sits on your local machine and provides a UNIX socket to which your sql client connects. The proxy has a private key which it uses to authenticate to the cloud SQL server, so you don't have to white-list anything etc. etc.

Still with me?

Bernard Meurer

Yeah yeah, I'm following

DanielSank

Ok, well I set up the proxy on my laptop and it worked just fine.

Then I go into my office and try it on my desktop and it doesn't work.

Bernard Meurer

Hm, firewall?

DanielSank

5:21 AM

When you run the proxy, you give it a path where it should put the socket. I got an error message that the path I used was invalid.

Nah, not firewall.

This was an outgoing connection so I doubt our firewall would have anything to do with it anyway.

So I'm wondering why the path would be invalid.

The path I was using was:

/usr/google/home/danielsank/src/pyle/junctiondb/cloudsql/

Should be fine.

Nothing weird there. No funky characters.

I grabbed my coworder for help. I told him I thought it could be an issue with the length of the path.

Bernard Meurer

Yeah that's a normal path

DanielSank

He pointed out that there's really no way that could be it, but he suggested it could have something to do with permissions.

Bernard Meurer

UNIX limit is huge for path size

DanielSank

In particular, he was worried that because we were going through /, maybe permissions were messed up.

So, in the command line flag for the socket location, he shorted it to:
./cloudsql/
which was ok because our current working directory was /usr/google/home/danielsank/src/pyle/junctiondb already.

It worked.

Now...

He thought that suggested a permission issue, but I wasn't convinced.

The thing is, having a path originate in / can't mess up permissions if the actual full location is accessible.

I mentioned length again.

Bernard Meurer

Yeah that's right

DanielSank

5:26 AM

So, we looked at the system log.

We noticed that the error was coming from where the proxy was invoking the UNIX "bind" command.

(bind is how you associate a program with a socket)

It was at the call to bind that the "invalid path" was originating.

Bernard Meurer

Wat

How?

DanielSank

So, my coworker pulled up the documentation for Linux's bind call.

We noticed that the path argument was defined as a string buffer of length 108.

(Why is it 108, I have no idea)

Bernard Meurer

WAT

hahahaha

DanielSank

So we checked the length of the path we were using. It was only 64 or something.

...but then my coworker noticed that the name of the socket was really long, and the path plus the name of the socket was 120 characters.

So I got my laptop, made the directory name long enough to make the full path and socket name 109 characters and tried it.

Same error :)

Bernard Meurer

What on earth

DanielSank

5:30 AM

So then I went home.

@BernardMeurer Yeah.

So now I'm learning how Google bug reporting works.

Bernard Meurer

linux.die.net/man/7/unix

Ah

There's a specific path size just for bloody UNIX sockets

talk about obscure

DanielSank

Yup.

Bernard Meurer

#define UNIX_PATH_MAX 108

DanielSank

Yep.

Bernard Meurer

Good thing you aren't me

DanielSank

5:34 AM

Why?

(and what does Unix(7) mean?)

Bernard Meurer

I'd have opened the goddamn source and changed that define and recompiled the whole damn thing

DanielSank

@BernardMeurer o_O

Bernard Meurer

ain't no one stopping me from using my paths how I want

My system, my rules

DanielSank

LOLOLOL

You'd recompile your whole OS?

Jeezus

Bernard Meurer

@DanielSank just the bind tool

DanielSank

5:35 AM

@BernardMeurer Yeahh that's probably used in a lot of the OS.

Bernard Meurer

I have done this before you know, for silly reasons like this

linuxfromscratch.org

DanielSank

That's kernel level stuff, isn't it?

Bernard Meurer

papa bless that site

@DanielSank yep, that's why I don't have a working computer half the time

DanielSank

@BernardMeurer Dude, that's what it takes to become a true master.

user54412

well, a light enough distro takes only a few seconds to compile

DanielSank

5:36 AM

Good for you.

@ChrisWhite Scroll up and read my bug story. You'll like it.

user54412

I've been reading :)

DanielSank

oh ok

Bernard Meurer

@DanielSank I've been told if your computer survives the process of learning linux in its completude you haven't really succeeded :p

DanielSank

completude?

Bernard Meurer

@DanielSank Yes, I made up that word

Chrome tried to correct me but I'm in a rebel mood

DanielSank

5:38 AM

@BernardMeurer Wait, but I found it in a French Wiktionary site.

complétude

== Français == === Étymologie === (1943) Mot composé de complet et -tude. === Nom commun === complétude \kɔ̃.ple.tyd\ féminin (Topologie) Qualité d'un espace métrique complet, ou d'un espace vectoriel normé complet. (Rare) Qualité de ce qui est complet, exhaustif. La portée des données du point de vue géographique et ethnique, la qualité des statistiques — c’est-à-dire leur degré de complétude. — (Annuaire démographique, United Nations Statistical Office, 2000) Afin d’évaluer le degré de complétude des données sur la mortalité enregistrées dans les pays dotés de registres d’état civil,...

wtf

Bernard Meurer

See, I could be french

@ChrisWhite Tiny core?

OH OH OH

I got a terminal to run on my kindle

user54412

@BernardMeurer should I feel insulted? :p

Bernard Meurer

@ChrisWhite Hahaha, no no, that's a very small distro

DanielSank

@BernardMeurer wtf

user54412

I actually run gentoo with minimal bloat

DanielSank

5:42 AM

How?

Bernard Meurer

@DanielSank Turns out that thing already sort of runs some weird UNIX-like thing

I just had to root it

DanielSank

@BernardMeurer ...

Bernard Meurer

I wanted to change the screensavers, the normal OS wouldn't let me

DanielSank

@BernardMeurer I can tell you are destined for greatness because you share a very special characteristic with one of the greatest programmers of all time.

Bernard Meurer

8 mins ago, by Bernard Meurer

My system, my rules

DanielSank

5:43 AM

Bernard Meurer

@DanielSank HAHAHAHAHA

DanielSank

When I read your message in the kindle's terminal, it reminded me of this image.

Bernard Meurer

Yeah I take inspiration from Torvalds, in that picture he's saying "Fuck you Nvidia"

The words of a true poet

DanielSank

@BernardMeurer lololol

Yes.

user54412

pity whichever of your minions breaks userspace, or even logs a short commit message in the repo

DanielSank

5:46 AM

@ChrisWhite Hmmm?

Bernard Meurer

@ChrisWhite Hm? I ain't got no minions

user54412

@DanielSank typical one of his messages

user54412

@BernardMeurer yet

Bernard Meurer

AAAAAH I've read that one

Love it

Mauro did make a shitty patch though

DanielSank

ehhhh, that tone is not helpful at all.

There's another good one from the git list where Linus 'explains' why he doesn't like C++.

Bernard Meurer

5:51 AM

@DanielSank Torvalds is known to be unhelpfully angry at times. Mauro's previous email wasn't good either, and his patch was filled with bad ideas

DanielSank

Before reading this, I should mention the in between Dmitry's and Linus's emails, there was someone who popped in and said something like "get your flame-proof undies on".

lwn.net/Articles/249460

Bernard Meurer

Hahahahaha

> Quite frankly, even if
the choice of C were to do *nothing* but keep the C++ programmers out,
that in itself would be a huge reason to use C.

LOLOLOLOLOLOLOL

Well he is right about Boost

my experience with it wasn't good

GMP treated me far better

DanielSank

> So I'm sorry, but for something like git, where efficiency was a primary objective, the "advantages" of C++ is just a huge mistake. The fact that we also piss off people who cannot see that is just a big additional advantage.

Bernard Meurer

OH @ChrisWhite I got it working btw :)

Torvalds is a savage in these emails

It's the nerd UFC

user54412

@BernardMeurer And imagine that was 2007. Boost has probably gotten a lot better since then? Though C++ has certainly only gotten worse.

Bernard Meurer

5:56 AM

I didn't like the introduction of automatic typing

It's like fitting a square into a round hole

user54412

that was definitely a colossal mistake

user54412

if you can't figure out the type of your object in such a strongly typed language, then you've already lost, and no compiler tricks will save you

DanielSank

@BernardMeurer really?

Bernard Meurer

@DanielSank Uhum, because of what CW just said

DanielSank

@ChrisWhite In Scala I find it nice that I don't have to always write out the types.

user54412

5:58 AM

depends on language -- Python doesn't need explicit types either

DanielSank

Scala has a real type system. It's nothing like python.

Bernard Meurer

@ChrisWhite See if this looks cleaner, and this specially

DanielSank

I see auto typing as a readability bonus, but I can see that paying for that with compiler complexity could be bad.

Bernard Meurer

OH DAN

I FORGOT

user54412

but in C(++), either you can deduce the type yourself and using auto thus passes on the opportunity to have self-documenting code, or else you can't and you have no business proceeding until you can

Bernard Meurer

5:59 AM

I got into the next phase of that scholarships program :)

Have an interview monday in São Paulo

DanielSank

@ChrisWhite My point is that it can be really annoying to read all the type tags, particularly in cases where they simply are not helping you.

int x = 4;
var y = 4/3;

Actually that's a crappy example.

~sigh~

Bernard Meurer

Get rekt

DanielSank

@BernardMeurer Whatever.

@BernardMeurer Nice.

Bernard Meurer

I do get your point though

DanielSank

@ChrisWhite That's a really unfair comparison.

Bernard Meurer

6:01 AM

I just don't think if fits well into the C++ model

user54412

@BernardMeurer Much better. I'll even forgive your atrocious bracing style because at least it's consistent. ;)

DanielSank

Python box types everything. That's not the same as deducing types which are statically deducible.

Bernard Meurer

@ChrisWhite What's with my bracing :O

DanielSank

brace yourselves...

::dodges tomatoes::

user54412

@BernardMeurer That's, like, the common style everyone uses. You populist.

Bernard Meurer

6:02 AM

Hahahaha

I laughed :p

@ChrisWhite I used to break line before the brace but I got a ~lot~ of shit for it

so I changed my personality

;-;

user54412

As a physicist I refuse to use asymmetrical braces. If you won't let me put the opening brace on its own line, then you can be sure I'm putting the closing brace on the same line as the last statement.

user54412

And then we'll just have python style, with unnecessary end-of-line characters

user54412

btw, google-overlord says don't parenthesize return expressions

Bernard Meurer

@ChrisWhite Look at the code again :)

I was in doubt what the norm was for returns, lemme change it

user54412

@BernardMeurer that probably is going to offend more people than it makes happy :p

DanielSank

6:08 AM

Is there not a universal law about braces in C++?

You can talk crap about python all you want, but at least there's a well established intergalactic law about style.

Bernard Meurer

@ChrisWhite I'm in a rebel mood today

Crap, my kindle just beat me at chess

I'm a disgrace

@DanielSank Python is my favorite language O.o

C++ is just an affair

DanielSank

@BernardMeurer That was directed at Monseur Le White.

Bernard Meurer

@DanielSank I haz question

SO

DanielSank

@BernardMeurer I has answer, maybe.

Bernard Meurer

I'm writing my own testing solution for this euler totient project

user54412

6:13 AM

@DanielSank I have nothing against python. In fact I'd probably like C with forced indentation rather than braces.

Bernard Meurer

and I wanted to check if all values for n=1...500 will be correct

so I stored them all in a .csv file

and the test utility parses the file and tests them

is that a bad solution?

DanielSank

@BernardMeurer No that's fine.

Bernard Meurer

I though it was cool because I can add more test cases by just adding columns to my CSV

DanielSank

@BernardMeurer Yes.

When I have a situation like this, I first try to think of a way to model the data.

Lots of my code comes with functions specifically written for that purpose.

For example, if I am writing a function to fit a parabola to some data which is parabolic but with noise, part of the module has a function that generates test data which is parabolic with noise.

However, if you're in a situation where generating the test data is expensive or impossible, it's totally correct to just store the test data in the repo.

Nothing wrong with that.

Slereah

67

Q: How to include a Lenny face in a LaTeX document? e.g. ( ͡° ͜ʖ ͡°)

How would one include a "classic" Lenny face in a LaTeX document? A paper is due tomorrow and I am curious. e.g. ( ͡° ͜ʖ ͡°) I have attempted to use the verbatim environment but it has proved fruitless.

user54412

6:15 AM

@BernardMeurer I will say that when I use standalone opening braces, I simultaneously don't brace single statements. That way the style isn't too open and spread out.

DanielSank

@Slereah Oh for heaven's sake...

Bernard Meurer

@ChrisWhite I always brace single statements

user54412

This annoys a lot of people, but for anyone familiar with python it should read pretty fluidly.

Bernard Meurer

@DanielSank Good :D

user54412

I'm always told things like "but what if you add another statement and forget to add braces" to which I reply "so do you code in Notepad or Word?"

user54412

6:17 AM

seriously, of the many thousands of bugs I've made, my text editor has never let me do silly things like that

Bernard Meurer

@ChrisWhite You're underestimating people's stupidity

Also, fuck it, I'm running x-server on this TI calculator

I want to play doom

DanielSank

@BernardMeurer ...why?

Bernard Meurer

@DanielSank I only got linux to boot on it, I want to see if I can make X work :D

then play doom

user54412

I'm starting to wonder if Brazil even has modern computing equipment, or if you're forced to hack wristwatches and alarm clocks to access the internet :p

DanielSank

heheheh

Or @BernardMeurer actually lives in North Korea.

Bernard Meurer

6:21 AM

@ChrisWhite I think the only way I'll be decent at Linux is by working on these hardcore things :p

it forces you to understand the system

Makes me wonder if I can run linux on a watch now that you mentioned it

nice project idea

DanielSank

You're crazy.

Bernard Meurer

AWWWWWWW YEAAAAAH

@DanielSank You don't know the half of it :)

DanielSank

@BernardMeurer Yes I do. You impress girls with probability calculations.

...and apparently it works.

You truly are insane.

Slereah

Where is the keyboard

Bernard Meurer

@DanielSank Hey, that was a cute move man, I showed her that we meeting was a miracle

I swear she liked it

DanielSank

6:26 AM

@BernardMeurer I know! That's the crazy part.

Bernard Meurer

@DanielSank Dude, try it on your girl and see for yourself!

You think I'm insane now? In my sophomore year I asked the school's treasury for money to build a jet-powered bicycle for the science fair

They denied my funding though

DanielSank

@BernardMeurer My girl is an astrophysicist and makes my stats-fu look like a small child playing with finger paint.

@BernardMeurer What did you think would happen?

Bernard Meurer

@DanielSank Ask her what the odds are of the meeting happening twice, I couldn't respond to that

DanielSank

(I made a small rocket powered car once. The whole neighborhood came out to watch me launch it. It hit the curb and exploded. It was cool)

2

Bernard Meurer

@DanielSank I thought I'd build a damn jet engine

DanielSank

6:29 AM

@BernardMeurer To be honest, she'd think about it, and either become enraged and quit or become totally obsessed and not give up until she got the answer.

Bernard Meurer

I had the blueprints and all

@DanielSank Hmmmm, this just gave me an idea

DanielSank

@BernardMeurer uh oh

Slereah

>stats

snore

Bernard Meurer

@DanielSank Whenever I get married remind me of this probability thing, I have a plan to get many husband points

DanielSank

@BernardMeurer k

@Slereah I was never interested in statistics until I learned statistical mechanics.

Slereah

6:33 AM

Well here's the problem

DanielSank

Then a few years ago I bought a book of fifty hard probability problems.

Slereah

Don't learn statistical mechanics

DanielSank

Now I really like it.

@Slereah wat?

Slereah

Also boring

Bernard Meurer

Brazil right now:

bool impeachment(bool vote){
vote=true ? impeachment(false) : impeachment(true);
}

DanielSank

6:34 AM

@Slereah That's easy to say until you learn it. Then you'll see how cool it is.

Bernard Meurer

Well I'll be off to bed, almost 4 in the morning

Nighty @DanielSank @ChrisWhite

DanielSank

@BernardMeurer ciao

Slereah

I've learned statistical mech

DanielSank

I'm off too.

Slereah

I've had a class and all

user507974

7:07 AM

@Slereah a class? you only ever took 1

Slereah

7:22 AM

Well yes

Because it's boring

user507974

7:44 AM

@Slereah what?

boring?

it like describes EVERYTHING

Slereah

well no

QFT describes everything

Statistical mech approximates everything

It's for people too lazy to find exact solutions

user507974

@Slereah a vanishingly small approximation often

@Slereah and quantum at least to the level I've done it seems like a far worse approximation

@Slereah how is qft different

sure stat mech is a pain in the ass to work with sometimes but basically it underpins almost every multibody problem, and dont you generally learn important qft tools like renormalization groups originally in a classical statistical mechanic treatment

Slereah

Well the real thing to do is

solve the QFT equations for the entire universe

No need to assume any thermodynamical approximation

it's not that hard jeez

Just solve the universe

user507974

i may not be well versed in qft but im pretty sure its not as easy as you put it

Slereah

Just solve this

It's not rocket science

user507974

7:52 AM

@Slereah still, that doesn't just give you an equation you can plug an IV into and tada the universe pops out exactly as we predict

you dont get each particle in each position at each point in time

Slereah

Well that is where you are wrong

user507974

(or whatever the appropriate basis is for qft)

Slereah

There are no particles in QFT

You just nead the field configuration at a time $t$

user507974

@Slereah and the field evolves completely deterministically?

Slereah

The wavefunction does, yes

The measurement maybe not so much

user507974

7:55 AM

@Slereah well the measurement is part of the universe isnt it

Slereah

But then you can calculate the amplitude for that

Easy peasy lemon squeezy

user507974

@Slereah you still cant predict the measurement before you do it

or at least cant make a single prediction of what exactly it is

Slereah

Well yes but then that is even more against statistical mechanics

You're just gonna get results on very broad and deterministic variables

Instead of actual measurement probabilities

You're gonna get ENTROPY

Entropy is bullshit man

You don't need entropy

user507974

@Slereah well so you decided to rag on stat mech for not describing the exact dynamics of a particle, meanwhile you are putting on a pedestal a system that completely ditches the idea of even trying to describe an individual particle and converts the universe into a field

Slereah

Just because there are more field configurations for a given variable doesn't mean it's gonna happen!

user507974

7:58 AM

or a bunch of fields

Slereah

ALSO

Obviously, what you say only applies if you adhere to the Copenhagen interpretation

If we go by Bohmian quantum theory

Then everything is deterministic

no problem

user507974