The Nineteenth Byte

Yes, mine had to also. I wrote an example bot in Java to show how to do it, but basically it boiled down to checking with the available() method in a loop with sleep.

It's a bit hackish, honestly, but it worked well.

Nathan Merrill

I just have a try catch

since exceptions don't really cause a slowdown in python like they do in compiled languages

Geobits

6:17 AM

So what's the problem?

Just the added complexity for the bots?

Nathan Merrill

no, there's a real problem with the piping

I'm not getting messages through, and not getting responses either

Geobits

So when you send something to the bot, it isn't received, and vice versa?

Nathan Merrill

yeah

the bots getting run though

so at least one good thing :P

Geobits

How are you writing to the bot, exactly?

Nathan Merrill

piping

so, in python, the bot will do raw_input()

Geobits

6:21 AM

I mean piping it how, with what function(s)

Nathan Merrill

its a bunch of code...but in essence I call subprocess.Popen() which returns an object with stdout and stdin. If I just try to read them here and now, they will block, so I put them in a Synchronized queue

Geobits

You shouldn't need a queue to write to them, though.

Nathan Merrill

in case the bot isn't currently accepting input, then, yes I do

Geobits

Why wouldn't it be accepting input?

Nathan Merrill

it's processing

basicall, writing to the pipe will block me until the other bot does a readline()

if there is a .1 second sleep, I have to wait for that .1 second

but with Queues, I don't

Geobits

6:27 AM

Right, but how long is the bot going to be processing? They shouldn't be doing much else if they have no input, right?

Nathan Merrill

supposedly. I'm still allowing myself to get locked up though

Geobits

Only if a bot is locked up. That's why mine had a strict 100ms timer (or somewhere around there).

If it took too long, the program just skipped its turn and continued.

Nathan Merrill

I can't do that without threading

hence the queues

Geobits

Is there really no way to? That was admittedly the ugliest part of my code, but it wasn't impossible.

Nathan Merrill

nope. I can throw an exception with Ctrl-C, but python will just sit there

Geobits

6:31 AM

So really this is only a problem if/when a bot misbehaves?

Nathan Merrill

yes

well, kind of

it also improves speed

Geobits

"If a bot causes my program to block/lock, it's disqualified"

Nathan Merrill

...here's the thing...every program will lock mine

if it sleeps for .1 second, then prints, I'm locked for that period

Geobits

Well... yes, but that's normal.

Nathan Merrill

times that by 50...that's 5 whole seconds of wait time

queues, it becomes 0

because the queues are on a separate thread, accepting all input

Geobits

6:33 AM

Except that you can't get queues working :/

Nathan Merrill

I'll get it to work :P

Geobits

Ok, so assuming you get the controller wait time to 0 with queues, what else would it be doing if you don't have responses from bots yet?

Nathan Merrill

its blocked

on its own thread

wait...what's defined as "it"

Geobits

Gotcha

Nathan Merrill

the controller?

Geobits

6:35 AM

Yes

Nathan Merrill

oh, the controller will be doing logic

so, the queues have a nice timeout parameter

I say "if there isn't any input within X seconds, throw an error"

and if it does, I output the name of the bot that isn't working

and then continue along

Geobits

Ok, I did basically the same thing but not in a queue.

Nathan Merrill

so, there's still a timeout for unresponsive bots, otherwise, I just grab the output, and work with it

Geobits

I don't know python queues at all, though, so if your problem is specific to the threading issues it creates (as it seems to be), I'm not sure how much help I can be.

Nathan Merrill

well, I'm really just following this post: stackoverflow.com/questions/375427/…

and making it work for output too

Geobits

6:40 AM

Doing it sequentially was quite fast for me, though. 40k rounds in under 20 minutes, and the vast majority of that was controller logic. Pure I/O was probably 25% or less.

That was with 10 entries, YMMV.

Either way, it was absolutely much, much faster than doing a process per turn like many koth.

Nathan Merrill

aha!

Geobits

Eureka...?

Nathan Merrill

sort of

I didn't test getting STDOUT properly

apparently, it works just fine

Geobits

Nice

Nathan Merrill

its STDIN that is the problem

Geobits

6:43 AM

On the controller or bot side?

Nathan Merrill

bot

bot can't read

Geobits

No Bot Left Behind

When you say "can't read",do you mean it reads and there's nothing there, it throws an error, what?

Nathan Merrill

it throws an EOF error :P

it just isn't getting anything

Geobits

using raw_input, right?

Nathan Merrill

yeah

Geobits

6:47 AM

so when you catch the error, you loop back after a short wait and try again, right?

Nathan Merrill

yeah

Geobits

And it never gets anything from the controller?

Nathan Merrill

nope

Geobits

Well I'm not sure how that could be the bot's fault then, that part should be straightforward.

I've gotta head out for now, though. Good luck!

Nathan Merrill

bye!

hmmm...I'm wondering whether it would be easier to make this a non-stay alive program

luser droog

8:11 AM

Can you make a computer that runs on steam?

I read somewhere that you can make hydraulic transistors.

You wouldn't want paper cards for input, because they'd get too soggy.

Maybe some kind of metal chain where the links can be in different configurations, or different types.

Instead of a keypunch, you'd use something more like a linotype.

Heavy metal programming. :/

Closer to reality, I want to design and build an espresso machine with a built-in player-calliope. I'm not entirely sure about the design of the player part; either punched cards or discs. But if you've got a source of steam, why not have a built-in calliope?? And to that question, I have not found a satisfactory answer.

Martin Büttner

9:27 AM

@luserdroog Probably... you might just get a problem with miniaturisation ;)

@Doorknob But there has been a meta post and it's not like it really accomplished anything

@Sparr not bad!

luser droog

10

Q: What would the replicator do if Picard ordered "Hot Earl Grey tea"?

Is there a particular required sequence used for replicator request descriptions? Is there a reason Captain Picard orders his tea, "tea, Earl Grey, hot" aka "noun, descriptor, descriptor" and not "descriptor, descriptor, noun" or "descriptor, noun, descriptor" (in case he missed a descriptor)?...

Martin Büttner

@NathanMerrill If it's a good puzzle with multiple solutions there are still ways to rank them, by code length, upvotes, etc. But in general yes, that is a big problem with puzzles within the SE format. How does Puzzling.SE do it? Do they actually pose puzzles or just discuss them?

^ @Doorknob (you're a mod there too, aren't you?)

@luserdroog Ha, I saw that in the HNQ on my phone this morning and was going to check it out now :D

Doorknob

@MartinBüttner I'm not even sure what the process is for changing a site's name, but I'll try to figure that out

luser droog

There's one really good answer, with quotes.

Doorknob

@MartinBüttner Yep; it's not really the same as PPCG there in that there's no objective winning criterion; it's more similar to "regular" SE sites. Although people still do post puzzles to solve, but they're mostly "first correct answer wins" (or sometimes "most through answer")

Martin Büttner

9:35 AM

Ah I see

Eoin Campbell

10:56 AM

New Results up @ codegolf.stackexchange.com/questions/35079/…

@MartinBüttner sorry dude... new guys are slipping you down

:-)

Martin Büttner

@EoinCampbell :(

and that pony just can't be beaten :D

I think I'm still up for the "Concise" award though :P

Eoin Campbell

looks like it. gazzr gave it a run for its money. but was beaten 43-37. if that result had gone the other way the tourny would hae been a 63/63 tie

Martin Büttner

11:50 AM

@VisualMelon would you mind sending me your circuit solver, so I've got a program to test my controller with?

Nathan Merrill

2:06 PM

@MartinBüttner Puzzling.SE isn't a competition. Answering the questions if for the benefit of the questioner

Martin Büttner

@NathanMerrill Ah yeah, I wasn't sure about that. Thanks

Sparr

3:31 PM

@MartinBüttner my perfect 1-256 is a lot slower than yours, and I'd like to improve that just because it would mean getting more data for the bigger numbers.

VisualMelon

@MartinBüttner visualmelon.onl/dominoPrinter.cs

Sparr

or maybe I'll just ignore it. I've got a bad habit of only doing the challenges without much participation

Martin Büttner

@VisualMelon cheers, how do I compile it from the command line?

@Sparr happens to me, too

VisualMelon

csc dominoPrinter.cs ought to do it

it write out a lot of rubbish to stderr, do you might need to re-direct that somewhere

(and don't give it the 4-bit parity checker or 6-input example if you don't want to wait forever)

Martin Büttner

3:49 PM

wow it took me ages to find csc

VisualMelon

oh, sorry

Martin Büttner

no problem

are you expecting the input on STDIN?

it's going to come in via argv actually

VisualMelon

I am indeed expecting from STDIN

I can change that now if you want

Martin Büttner

if you don't mind

VisualMelon

wow, don't know how I missed that

Martin Büttner

3:55 PM

it wasn't as clear before

the spec said "command line argument" but the example had a line break between M N and f

so it looked more like STDIN

VisualMelon

re-uploaded .cs to use command line args

Martin Büttner

cheers

would you mind renaming the file to VisualMelon.cs, so I don't have to rename it every time? :D (just in case I need to redownload again)

VisualMelon

sure ;)

Martin Büttner

4:33 PM

gaaaah, c++ and it's bloody broken modulo

VisualMelon

I did the 24-hour IEEXtreme competition this year

and one of the questions requires some dodgy Modulo behavior

Nathan Merrill

...c++'s modulo works...?

VisualMelon

that was most annoying, everyone using Java seemed to complete it in no time, and we just skipped it rather than trying to reverse the behaviour from the examples

Nathan Merrill

what is broken about it?

Martin Büttner

negative numbers...

give negative results

how could that ever be useful

Nathan Merrill

4:36 PM

So...-9 mod 4 = -1 or -3?

Martin Büttner

it's -1

Nathan Merrill

what happens in Java? 3?

Martin Büttner

I don't know about Java, but I have never had a use case where I wanted the result to be -1 instead of 3

VisualMelon

according to docs, Java will do that same

Sparr

4:52 PM

if -9 mod 4 == 3 then what is -9 / 4?

VisualMelon

@Sparr hadn't thought of that, but -9 % -4 is still -1

Martin Büttner

5:07 PM

0

A: Tweetable Mathematical Art

Random painter char red_fn(int i,int j){ #define r(n)(rand()%n) static char c[1024][1024];return!c[i][j]?c[i][j]=!r(999)?r(256):red_fn((i+r(2))%1024,(j+r(2))%1024):c[i][j]; } char green_fn(int i,int j){ static char c[1024][1024];return!c[i][j]?c[i][j]=!r(999)?r(256):green_fn((i+r(2))%1...

@Sparr I believe it's -2 such that (m / n)*n + (m % n) = m

VisualMelon

that's beautiful!

Martin Büttner

thanks :)

Sparr

images won't load for me right now, but I'll look at it later

Nathan Merrill

I got messaging between persistent programs working ^^

Sparr

@MartinBüttner the "random adjacent pixel" seems to favor northwest and/or southeast pixels?

Martin Büttner

5:13 PM

@Sparr only southeast

that's the first of my bullet points at the end

I can only pick them at larger coordinates, because I don't have enough characters to handle underflow in red

Sparr

oh, sorry

also, are you relying on the array being initialized to zeros?

Martin Büttner

I'm currently trying out what happens if I fix it for green and blue

@Sparr yes

static storage

should be 0

okay if I can save one more character, I might be able to make it isotropic

Sparr

what is the format for the pixel art wrapper output?

MathPic?

Martin Büttner

ppm

is that what you mean?

Sparr

ahh, sure

gonna put an extension on the file so my system knows what to do with it :)

Martin Büttner

5:19 PM

imagemagick can convert it to png

Sparr

yeah

Martin Büttner

ha, found that one character

Sparr

would it be cheating to rename the functions?

so wasteful for them to be red_fn instead of r, etc

the challenge technically is just for the function bodies

Martin Büttner

i know, right?

there would be a lot of interesting possibilities, if we had a bit more leeway with the framework

like forward-declaring the three colour functions

Sparr

he says you can use another language

which makes me think that manipulating the framework would be reasonable

Martin Büttner

5:23 PM

yeah I thought about that as well

Sparr

as long as you aren't introducing additional functionality

something is wrong with my framework

I'm just having red_fn return i and green_fn return j and blue_fn return 0

that should give me a 2d black/red/green/yellow gradient, right?

Martin Büttner

yes

is it oscillating like crazy?

Sparr

I'm getting a grid of very tiny such gradients

Martin Büttner

there's some little endian/big endian trouble

Sparr

that would do it

Martin Büttner

5:27 PM

I switched to using chars instead of shorts and reduced the maximum colour to 255

btw I added isotropic random walk to my post

I prefer the diagonal strokes though

New Sandboxed Posts

0

A: Proposed Question Sandbox - Mark XIV

Generate a random work of art We've all seen minimalist modern art at museums and thought, I bet I could do that. Well, now's your chance. Your challenge is to (pseudo-)randomly generate an image in the style of Piet Mondrian's Composition paintings, but make it look like a real painting, not a ...

Martin Büttner

also wow, rand is SUCH a band PRNG... if you just generate all colours randomly you can see so much structure in the picture

Sparr

void pixel_write(int i, int j){
static unsigned char color[6];
unsigned short red = red_fn(i,j)&DM1, blue = blue_fn(i,j)&DM1, green = green_fn(i,j)&DM1;
color[1] = red&255;
color[0] = red>>8;
color[3] = green&255;
color[2] = green>>8;
color[5] = blue&255;
color[4] = blue>>8;
fwrite(color, 1, 6, fp);
}

my rand() is pretty unstructured

Martin Büttner

ah that's a nice alternative, but I can't use it now because it would require one more character in my code :D

(to replace a 255 with 1023)

Sparr

1024 instead of 256 as a const?

yeah

there's virtually no structure in my random output

that's 1/4 of the output

Martin Büttner

5:35 PM

huh

Sparr

just rand(1024) for all three channels

err, r(1024) which is rand()%1024

Martin Büttner

huh

well I had r(256)

and got a lot of diffuse vertical lines

Sparr

I got vertical lines when imgur converted my full size image from png to jpg

no idea why that happened, and they weren't present or visible in my original

those dark lines are not present in the original image

Martin Büttner

ah no, I don't mean lines like that... they were more large scale patterns in a few places

VisualMelon

looks like carpet

Martin Büttner

5:38 PM

well, now that I finally have a submission I'm really happy about I can turn my back on that challenge

Now I'm almost tempted to implement rule 110 in that challenge, but 4 submissions are really enough :D

Sparr

5:56 PM

I'm thinking of doing an entry that does an arbitrary CA rule

or a randomly selected one

or overlaps three randomly selected ones

Martin Büttner

yeah exactly, something like that

Sparr

@MartinBüttner why !c[i][j]?foo:c[i][j] instead of c[i][j]?c[i][j]:foo ?

Martin Büttner

6:18 PM

@Sparr because you wouldn't gain a character from it

you'd need to replace the ! by a space

and this way around seemed to be the more logical way to write it for me

@githubphagocyte why does area 51 state that a proposal needs 40 followers in the definition phase if you need at least 80 followers (in the unachievable best case scenario) to fulfil the requirement regarding the questions?

githubphagocyte

@MartinBüttner I have wondered this myself. My only guess is that this provides several layers of targets to hit, so that we don't get put off by seeing how far away the end is. Lots of little journeys rather than one big one. Of course it doesn't help people who see through it...

@MartinBüttner if you were talking about the tweet art question when you said you couldn't spare a character to use 1023, you can use DIM for 1024 and DM1 for 1023, if that helps.

Martin Büttner

@githubphagocyte Ah, good point. I don't think I can be bothered to switch back to short right now ^^ (of course, that macro could still save a few characters in different places)

Sparr

6:59 PM

damn

have a entry for tweetable mathematical art at 166 bytes for one of the functions

I can fix this!

Martin Büttner

:)

wanna give us a preview here? :P

Sparr

it's just three random 8-bit CAs overlaid, one per color channel

cellular automata