The Nineteenth Byte

Martin Büttner

9:00 PM

is the acceleration constant?

feersum

what about ramming?

shooting bullets with other bullets!

PhiNotPi

@MartinBüttner Yes. The hovercraft fans are at a certain constant strength that changes upon command.

Geobits

Can I shoot the bullets in a curve? :)

feersum

I'm in favor of a physics-based challenge

Martin Büttner

@PhiNotPi Δx = (1/2 a Δt + v_0) Δt

Rainbolt

9:02 PM

Can I name my hovercraft Neo?

PhiNotPi

The problem is that there is angular acceleration and "linear" acceleration happening at the same time.

@Rainbolt sure

Martin Büttner

these should be independent of each other, provided the angular acceleration is applied about its centre of mass and the linear acceleration through its centre of mass

feersum

can you make someone go into a spin by shooting them?

Geobits

That would be sweet. Hit them off center and apply a rotation force.

Martin Büttner

that can also be decomposed exactly into torque and linear force

shouldn't be too hard with circles

I can look up the formulae in my lecture notes if you need them

PhiNotPi

9:05 PM

If there is a force of 4 on the right side and 1 on the left, the hovercraft will travel in a curve. How do I solve that?

I would appreciate the help.

Martin Büttner

oh, I see

well you can decompose both forces into torque and linear force, add up the linear forces and cancel the torques

Rainbolt

If there is a force of 4 on the right, and a force of 1 on the left, then there is a net force of 3 on the right.

Martin Büttner

@Rainbolt they're both in the same direction

Geobits

This reminds me of my way-too-involved boids sandbox I made years ago.

Martin Büttner

(think tangential forces)

Rainbolt

9:07 PM

@Martin He said one was on the left and one was on the right. They can't both be in the same direction...

Martin Büttner

@Rainbolt it's on the left... not to the left

imagine jet engines on both sides pointing in the same direction, the right one blasting more strongly than the left

Geobits

Or tank treads

Martin Büttner

(or fans, because that's what they are)

feersum

It would be funny if the turret cannot move and they only way to aim is to spin yourself (:

Geobits

I think that's the plan, no?

PhiNotPi

9:08 PM

Yeah, the plan is that you have to spin yourself.

feersum

great

PhiNotPi

I clarified the fan thing, I think. There's one on the left and one on the right, but they both face forward.

feersum

is there any 3-dimensional aspect or are the physics strictly 2D?

Geobits

I assume you want dual fans so it's easier to rotate quickly? Otherwise a single fan would be much simpler.

PhiNotPi

Strictly 2D as of right now.

Rainbolt

9:11 PM

@Geobits How would you rotate with a single fan? (winning)

Geobits

Make it aimable....

Rainbolt

You still can't rotate...

Martin Büttner

@Rainbolt if it's off-centre you can.

Geobits

You wouldn't be able to rotate in place, yes. It would give a loopy effect.

Rainbolt

Well yes, I was assuming the single fan would be placed at the center

Geobits

9:12 PM

Or off-center (at the back probably)

Think like a motor in a boat.

Martin Büttner

I think that would require too much physics of the participants though. it would be nice if they could just use the simple intuition of "blow stronger on the left to turn right"

feersum

I like the simplicity of two non-aimable fans, only on or off

Martin Büttner

the hardest thing is going to be detecting and resolving collisions.

with the acceleration I don't think it's easily possible to have exact collision detection

(you can only take small steps, and fix things once objects overlap)

PhiNotPi

I don't think exact collision detection is going to happen.

feersum

maybe we should find an established physics engine?

Geobits

9:14 PM

BTW, I'm going to steal whatever formula you end up using to play with my Ollie.

Martin Büttner

I've done it for this game prototype, but there was no acceleration

@Geobits oh, that's new

looks nice :)

@feersum box2d?

the equations involved should be simple enough though, especially since everything is circles

feersum

is it any good? does it handle circles?

Rainbolt

And there's only one force that actually moves the object. You can only go forward.

The other force just spins it in place

Martin Büttner

@feersum it's the one used by boxcar2d

@PhiNotPi you'll also need to decide if hovercrafts are sticky or slippery (on the edges), which is important when resolving angular momentum in collisions. of course, slippery is much simpler, because it means momentum can't be transferred.

sticky is more fun though :D

Rainbolt

I can already imagine the clown bot that spins as fast as possible and then transfers all of that spin to a poor opponent.

2

PhiNotPi

9:19 PM

Here's a simple scenario.... the hovercraft is accelerating "forward" but is already rotating. Acceleration in X is a*cos(θ), so velocity in X is the integral of that...

Sticky sounds fun.

Rainbolt

The hovercraft is always accelerating forward. You literally can't accelerate in any other direction

Martin Büttner

@PhiNotPi just make the time steps short enough so you don't have to worry about it

@Rainbolt yeah, but if your timestep is too long the meaning of forward changes significantly over the timestep

Geobits

@Rainbolt I assumed a -3 on left and +3 on right would rotate in place.

feersum

How many fps? should players be called every frame or skip some?

Rainbolt

@Geobits Oh, the thrusters can go backwards now?

PhiNotPi

9:21 PM

I was planning on backwards thrust.

Geobits

I inferred that from "Each fan has 8 power settings, from -3 to +4." May be wrong.

PhiNotPi

Not that it really matters.

Rainbolt

Oh, well that blows my theory

PhiNotPi

I'm not set on the -3 to +4, at all.

Geobits

Makes for a better clown car, though, just start a bit forward, then max spin :D

PhiNotPi

9:23 PM

I was thinking that, without a negative setting, it would be hard to stop turning.

feersum

turning is fun..why would you want to stop turning :D

PhiNotPi

Okay, when both fans are accelerating at the same rate, but the hovercraft is spinning, the resulting position formula isn't that bad.

I fed "integrate integrate acos(bt) dt dt" into Wollfram alpha, where "a" is the forward acceleration, and "b" is angular velocity, and "t" is time.

integral integral a cos(b t) dt dt = -(a cos(b t))/b^2+c_1 t+c_2 gives you the x-position as a function of time.

feersum

there's no need to integrate anything in a numerical mechanics simulation

PhiNotPi

I wanted to see if things can be solved exactly.

feersum

you simply update the velocity/acceleration each time step

with some formula such as RK4

Martin Büttner

9:34 PM

that's probably way more than you need

feersum

it's super easy to code though

Martin Büttner

RK2 will probably suffice, if not even a simple difference method, provided you make sufficiently small time steps.

PhiNotPi

That is what I may end up doing...

feersum

that could be better, if controllers are not activated every frame

I would if there should be any friction/drag, to prevent clownbots from having unbounded angular momentum

Rainbolt

If you move players in time steps, spinning ridiculously fast doesn't provide any advantage anyway.

PhiNotPi

9:39 PM

There will be time steps, but the game won't be turn-based.

feersum

@PhiNotPi What do you mean?

Rainbolt

If I can make a full rotation every time step, that's exactly the same as making two full rotations every time step, or a hundred rotations every time step.

Regardless of whether you are spinning ungodly fast or just one rotation per step, it's all the same right?

New Sandboxed Posts

0

A: Sandbox for Proposed Challenges

Find the direction of the Bicycle (code challenge) There is a trace of a bicycle in the snow, but you can only see the two lines of the two wheels. Your goal is to find out which direction the bycicle did go. Task Write a program that takes an black and white image as input and calculates whic...

feersum

not in a collision

PhiNotPi

Turn-based would mean that, before doing a time step, you wait for each player to give a response. I was planning on making it so that, if the player hasn't given any output, that the time step goes on anyways.

Rainbolt

9:43 PM

He hasn't said anything about sticky collisions yet

feersum

That seems like a nightmare scenario for coding the controller

pretty sure asynchronous communication is not something you want to go for

PhiNotPi

I think I will have some "minimal" friction to prevent bots from getting up to speeds where it makes an entire revolution per time step.

Why do you think asynchronous communication will be bad?

feersum

among many other things, bots can hog CPU from each other

Rainbolt

@PhiNotPi Why do you want to prevent bots from reaching angular velocities above a certain level? I was just trying to explain why it wouldn't make a difference.

Oh, I see. I just read that part of the challenge.

I didn't realize shots fired were instantaneous

PhiNotPi

The shots might not be instantaneous.

Rainbolt

9:47 PM

Seems like the obvious solution is to make shots fired an actual projectile

PhiNotPi

I haven't decided.

Rainbolt

I guess it would make the engine move slower if it had to render projectiles too

PhiNotPi

How fast should the projectile go?

Any ideas?

Rainbolt

Is there a board size?

PhiNotPi

Nothing that isn't completely arbitrary.

I picked a radius 400 circle.

Rainbolt

9:50 PM

Starting to think that maybe limiting velocity is the right way to go. I don't know how you would possibly hit anything that is moving at insane speeds.

feersum

with a limited board size, you can't reach an insane linear speed without hitting the edges

New Sandboxed Posts

0

A: Sandbox for Proposed Challenges

Every number is interesting We know that every number is interesting but how? You should write a program or function which: takes a list of N positive integers (>0 and <2^31) outputs N lines each of them showing how the corresponding input number is interesting is not longer than 1024 bytes u...

Rainbolt

I hate that NetBeans recommends turning this
        for (ArrayList<Tile> row : tiles) {
            for (Tile tile : row) {
                sb.append(tile.hasPlayer() ? tile.getPlayer().getTeamId() : " ");
            }
            sb.deleteCharAt(sb.lastIndexOf(";"));
        }
into this
        StringBuilder sb = new StringBuilder();
        tiles.stream().map((row) -> {
            row.stream().forEach((tile) -> {
                sb.append(tile.hasPlayer() ? tile.getPlayer().getTeamId() : " ");

feersum

What does it mean that Java recommends something?

Rainbolt

Sorry, I meant my Java IDE. Edited inline

feersum

9:56 PM

now with automated code bloating?

PhiNotPi

Since I was having trouble determining what size a game tick should be... After accelerating a distance of 800 (one edge to another), the bot can have a maximum speed of 40*sqrt(a).

So, I would want it so that a maximum speed bot won't experience a choppy motion.

Rainbolt

@feersum It looks like the benefit is that the operation I am applying to each item in the list could be executed in parallel, so using a functional stream allows that to happen. But I didn't understand it too well so I may be full of it.

feersum

That would be really surprising if it actually executes in parallel

does it really work?

PhiNotPi

With the hovercraft challenge, I guess I'm at the point where I need to write a lot of code and test out what works.

Sp3000

(Oh that's awkward - some languages have builtins for KS-test! Is it too late to add a no builtins rule?)

Rainbolt

10:06 PM

@feersum From the Javadoc: "All streams operations can execute either in serial or in parallel. The stream implementations in the JDK create serial streams unless parallelism is explicitly requested."

feersum

lol, that's what I thought

Rainbolt

You would think that if the only benefit is parallelism, that NetBeans would bloat my code to something that uses it.

feersum

better use stream..just in case someone runs it with a magic Java implementation that executes eveyrthing in parallel

Rainbolt

I'm still using NetBeans 7.4. Maybe NetBeans 8.X will behave a little better.

Peter Taylor

10:18 PM

@MartinBüttner Why not? Isn't it just solving a quadratic?

Martin Büttner

@PeterTaylor hm, possibly? it's basically finding the intersection of two parabolic tubes in 3D

yeah, sounds all quadratic to me

Peter Taylor

@feersum Quadrature is integration. Just done badly ;)

The thing which can get quite nasty with the hovercraft simulation is simultaneous collisions involving more than 2 objects.

feersum

yeah, this is why a preexisting physics engine should be considered, to make collisions someone else's problem;)

Peter Taylor

For Torquing! I solved it with the approach described in this paper

Martin Büttner

@PeterTaylor btw, I edited my list of ideas for randomness-based challenges to the Shuffle an Array sandbox post if you're interested (there are a couple more in there than I mentioned yesterday)

Rainbolt

10:38 PM

@MartinBüttner It looks to me like the close votes on this question are because the question is unclear, but the only objection I see in the comments is that the Enigma spec isn't in the challenge itself. Since when did "unclear" and "not self-contained" become synonymous?

Maybe you didn't vote, so I could be asking the wrong guy.

Martin Büttner

@Rainbolt I have no way of figuring out what I'm supposed to do without following the link. I'd call that unclear.

Rainbolt

Are the algorithms used by Enigma vague or ambiguous?

Martin Büttner

no, but I don't know what they are, and the question doesn't tell me

(actually I do, but that's beside the point)

that being said, I'm not sure they're completely unambiguous

I'm sure there must have been several models

the exact implementation he wants is "obviously" the one used by the simulator he linked, but I can't really reverse engineer that either

Rainbolt

You aren't sure the spec is unambiguous so you close voted as unclear?

Martin Büttner

I'm sure the spec isn't the challenge post, so I close voted as unclear

Rainbolt

10:42 PM

non-sequiter: The spec isn't in the challenge post. Therefore, the spec is unclear.

Martin Büttner

feel free to use your reopen vote if it gets closed

Rainbolt

I intend to

Doorknob

@Rainbolt How can a spec possibly be clear if it doesn't exist?

Rainbolt

@Doorknob Oh, the enigma spec doesn't exist?

I wasn't aware of that.

Martin Büttner

I don't see how a non-self-contained challenge is any different from an unclear one

Doorknob

10:43 PM

Not in the question, nope.

Rainbolt

That's not what I asked.

Doorknob

All Stack Exchange posts are expected to be self-contained.

If it's not directly in the post, as far as we're aware it doesn't exist.

feersum

There's no close option for "you were too lazy to write the question specification", so you gotta pick something.

Rainbolt

Well I don't see any votes to close because the post is not self contained.

I only see votes for unclear.

Doorknob

One of the main goals of Stack Exchange was to get rid of the torture of clicking through links and links and searching over and over again.

@Rainbolt There is no spec. Therefore, it's unclear.

Rainbolt

10:45 PM

I understand that. I'm just trying to understand the choice of vote here.

@Doorknob I literally just explained that that is a non-sequiter.

Doorknob

Why?

Martin Büttner

@Rainbolt I'm sorry, I must have overlooked the "Your post is not self-contained" close reason

Rainbolt

Custom reason...

Is it too hard to type a custom reason instead of using one that makes zero sense?

Doorknob

Why use a custom reason when unclear works just fine?

Direct quote: "Please clarify... add additional details... hard to tell exactly what you're asking."

Martin Büttner

@Rainbolt aren't there only custom reasons for off topic?

Rainbolt

10:48 PM

If he copies and pastes a well known algorithm, then all of you will suddenly find the question to be clearer, despite the fact that he has added zero information that didn't already exist?

Martin Büttner

btw, I found a link to end this discussion: en.wikipedia.org/wiki/Enigma_machine#History ... there have been different models

@Rainbolt as far as I'm concerned it didn't exist

if he hadn't included a link, would you then close as unclear?

Rainbolt

Have you ever written a challenge that communicated via TCP?

If so, did you define TCP?

Did you rigorously spec it out?

Martin Büttner

I don't think I have, and if I did, you were allowed to use built-ins that hide the details

Rainbolt

I would close it as unclear based solely on the fact that there have been different models. But you didn't bring that up until just now.

Martin Büttner

I doubt you can solve this challenge with CryptographyEngimaEncode["input"]`

feersum

10:50 PM

If there's a challenge requiring data from the Internet, it doesn't matter what protocol is used to get it

Martin Büttner

@Rainbolt I mentioned that it's probably the case, and the fact that I have to go to another website to find out if a challenge is a unclear makes it unclear already

you ignored my question though

would you close as unclear if there was no link at all?

Rainbolt

Whether or not there is a link is totally irrelevant.

Martin Büttner

@Rainbolt so you wouldn't? o.O

Rainbolt

I didn't say that

I now know that there are two enigma specs (according to you), so I would still close in either case.

Martin Büttner

well you wouldn't close it with a link. you say that the link is irrelevant. so if the link wasn't there, that means you wouldn't close it either.

Rainbolt

10:52 PM

..... What?

Martin Büttner

@Rainbolt ignore the fact that there are multiple models

Rainbolt

Ok. There is a single, well specced model out there on the Internet.

No, I would not vote to close.

No, I would not vote to close with a link.

No, I would not vote to close without a link.

Martin Büttner

okay

I think that's the point where we agree to disagree and cast our respective votes

Rainbolt

Why did you lead me down this stupid link question just to tell me that we should agree to disagree?

You make me answer stuff for no reason. And you remind me that I need to answer it too.

Martin Büttner

well because I thought you might actually close a question that doesn't contain any reference to the spec at all

in which case I might have had another argument

but this way I don't

Rainbolt

10:54 PM

I thought it was already understood that the link was not a reference to the spec.

Is the link a reference to the spec?

Martin Büttner

I don't know? I thought you considered the link a spec

it doesn't really matter to me

feersum

Would you close it in a box? Would you close it with a fox?

Martin Büttner

if there's no spec in the question I'll voite to close as unclear

Rainbolt

No. I considered "the enigma machine" a spec.

Martin Büttner

oh okay, fair enough

Rainbolt

10:58 PM

Assuming there is only one enigma machine, the spec is well known information that need not be repeated in the question. I originally assumed this stance because i had heard about some yearly Enigma machine competition for programming students.

Besides, the folks who cracked the original Enigma machine probably didn't vote to close Nazi Germany as "Unclear what you are asking".

Eric Tressler

does anyone here know javascript fairly well

Martin Büttner

@EricTressler Optimizer

Eric Tressler

Pardon me?

Martin Büttner

although I'm sure others can help you too

what's the question?

Eric Tressler

oh, him

okay, I've been putting some stuff in canvases on my site, but I want to avoid namespace conflicts

Martin Büttner

11:02 PM

oh, k... I'm not really familiar with canvases

Eric Tressler

var frac = new Object();

this is incidental to that

Martin Büttner

(I just use them as a tool to render WebGL in)

Eric Tressler

I do that, and then all of my "globals" are now frac.bla

Martin Büttner

oh okay

Eric Tressler

but i've also seen people wrap entire scripts in a

function () {

}

Is there anything particularly bad about the way I've been dealing with scoping?

Martin Büttner

11:03 PM

I guess it's just less convenient?

wrapping everything in a function is idiomatic afaik

Eric Tressler

I don't know how to initialize my program if all of the functions are inside of a function wrapper

if i have a function () {

Peter Taylor

@Rainbolt The folks who cracked the original Enigma machine did so before Nazi Germany existed.

Eric Tressler

containing, say, a function init() {

Martin Büttner

@EricTressler you return init

(from the wrapper)

Eric Tressler

so then in the script, I would let X = wrapper();

and then X();

Martin Büttner

11:05 PM

yep

although I don't see why calling the wrapper couldn't act as initialisation

Eric Tressler

I haven't really tried this other method yet, I just saw it while source diving a little

Martin Büttner

or do you want to run the wrapper before init for some reason?

Eric Tressler

no, no. I only want the wrapper to get function scoping

Martin Büttner

then just call init at the end of the wrapper

then you don't even need to expose that

Eric Tressler

I have some cases where I have 4 minor variants of the same program that I want embedded separately on a single page

and dealing with the scoping my way is pretty stupid

Martin Büttner

11:07 PM

if the minor variants can be parameterised easily, you could actually make that a single program, give the wrapper the parameter, and call the wrapper four times

Eric Tressler

I have like... Xv1.everything, then I do a search and replace by Xv2, etc.

nah, it can't

they're fundamentally different

Martin Büttner

k

Eric Tressler

plus they need their own animation stop/start controls

so they really need to be separate scripts

Martin Büttner

@EricTressler well they'd get their own controls

(if the wrapper instantiates new, local ones)

if you call it four times, you get 4 separate closures

Eric Tressler

ah. right. well, in this case it still wouldn't work

but that's good to know. i'll write my next program with a wrapper around it and get that working

Martin Büttner

11:09 PM

even if you have multiple wrappers, it would probably more convenient to be able to use the exact same names

(without prefixes)

Eric Tressler

yes

So I just got a job offer that conflicts with an active interview process with Google :|

Martin Büttner

:/

Eric Tressler

No idea how I'm going to deal with it yet. I just found out today

if they had just waited another 2 weeks, this would have been fine

Eric Tressler

11:24 PM

yeah, that worked

Sp3000

11:36 PM

@MartinBüttner Random chessboard sounds fun but also pretty insane. Are you counting only reachable positions in game? Also, are there pawn promotions?

Transcript for