any developer from cocos2d

I've got a some experience with c++ port of cocos2d

He's gone, you might want to add an @ his name

but I could tell what was his problem.

yeah, it's a very confusing question

from what I know that code should work, there has to be some other piece of code which contains main problem.

it doesn't help that he hasn't done a very good job explaining what the problem actually is, which i'm sure is a language barrier issue as opposed to anything else.

and he is being a bit of a jerk

@JoshPetrie hahah, Yeah

I was reading that question/answer too

probably smart

It's not entirely wrong

The HD question a little while back also made me laugh

C++ isn't usually the appropriate tool for any job, but it's what a lot of companies use.

I love @Tetrad's answer to the guy's question: "Yes... Unity supports HD"

my application at work features a 3000 pixel wide data grid..... can I claim I support super-HD?

@Jimmy yeah, I agree. If you're aiming for the highest end performance, you can go for C++, but it's rare for indie/hobby programmers to ever really need that

Go for ease over performance

well, in a high level language you get performance via ease of implementing smarter algorithms.

yeah, I guess you can... in a sense

heh, why do I always relate real-world problems to xkcd comics: xkcd.com/909 (worst case shopping)

The programmer in me claims that you only optimize the big-O of the worst case.

or that you profile your real-life flashlight usage first before deciding.

yeah, I agree. People that constantly preoptimize really bug me, to put it politely

A friend of mine does that all the time, and it's one of the reasons he and I have stopped talking about programming

yeah, coding should be optimized for legibility first. Although I find that legibility is quite subjective.

Or at least, that's what my coworkers tell me after reading my code.

:p

just as long as people aren't trying to be tricky in their code. I found this one piece where they had weaved an if/else with a #ifdef /#else and it was such a mess to pull apart so that I could add a single line

lol ouch

very ouch

at least they didn't also have gotos in there

I find upwards gotos in nested if/else nearly impossible to decipher.

downwards gotos are okay usually

not that I use them, but they are comprehensible.

:D I found the psudo-code I wrote myself:

oh confusing

And I needed to add a function call right infront of both foo(1) and bar(2), different calls for each

Notice that the "}else" hangs open for bar(2); if compiled with 32BIT_REGS

:( lovely

yeah, that was a fun day

So: No making puzzles in your code!

It gets real tricky when you combine those things with inlined assembly. :P

Is there anything like a TranslationScale matrix? Using a Translation matrix, I can offset the x, y, z values, but what if I want to scale the x, y, z values?

I think a scale matrix is a diagonal matrix

and can't you just combine the matrices?

just make sure to scale before transform

Matrix scaleMatrix = Matrix.CreateScale(this.scale, this.scale, 0.00f);
Matrix translationMatrix = Matrix.CreateTranslation(((FullView.Width - GameView.Width) / 2.00f) - this.world.ActiveRealm.RenderLocation.X, ((FullView.Height - GameView.Height) / 2.00f) - this.world.ActiveRealm.RenderLocation.Y, 0.00f);
Matrix matrix = scaleMatrix * translationMatrix;

That's what I'm currently doing.

The scaleMatrix will scale the texture size.

The translationMatrix will fix the x and y coordinates.

What I need, is to also scale the x and y coordinates.

But I don't know what kind of matrix scales the coordinates.

hmmm

if you apply the scale after the translation

wouldn't it also apply to the translation?

meaning: Matrix matrix = translationMatrix * scaleMatrix should apply the scale to the translation.

Does the exact same. Maybe matrixes can't scale coordinates, only offset them?

does your translationMatrix actually get applied?

Yea

:( hmm

@MindWorX How do these matrices work, exactly? Because scale (and rotation) is usually applies as multiplication, and translation as addition, so it's not easy to combine them into one matrix (or number, really).

well, here's my test code (using OpenTK vectors, I dont' have XNA handy)

var point = Vector4.One;
var scale = Matrix4.Scale(2f);
var translation = Matrix4.CreateTranslation(10f, 10f, 10f);

var matrix = scale * translation;
var matrix2 = translation * scale;
Console.WriteLine("Input: {0}", point);
Console.WriteLine("Result: {0}", Vector4.Transform(new Vector4(point), matrix));
Console.WriteLine("Result: {0}", Vector4.Transform(new Vector4(point), matrix2));

I must admit I don't know for sure, I've never used Matrices before. I just kinda wanted to test them a bit. But it seems I can't translate with a factor.

Input: (1, 1, 1, 1)
Result: (12, 12, 12, 1)
Result: (22, 22, 22, 1)

@MindWorX I didn't say it was impossible, just not easy. In 3D, that's done with dual quaternions (those can describe rotation, scale and translation in a single 8x8 matrix).

Actually ... in an 8-tuple, which gets modified into a special form of an 8x8 matrix.

Sounds like it'd be easier if I just make a method that does it for me instead.

@Jimmy: Can you check how the scale and translation matrices look like internally?

@MartinSojka : as you'd kind of expect

well, as I'd kind of expect

Translation
  (1, 0, 0, 0)
  (0, 1, 0, 0)
  (0, 0, 1, 0)
  (10, 10, 10, 1)
Scale
  (2, 0, 0, 0)
  (0, 2, 0, 0)
  (0, 0, 2, 0)
  (0, 0, 0, 1)
scale * Translation
  (2, 0, 0, 0)
  (0, 2, 0, 0)
  (0, 0, 2, 0)
  (10, 10, 10, 1)
Translation * scale
  (2, 0, 0, 0)
  (0, 2, 0, 0)
  (0, 0, 2, 0)
  (20, 20, 20, 1)

Yeah, 3x3 matrix for the scale/rotation part, extended by another row/column for translation (and projection).

translation * scale then translates, then scales (around the origin), scale * translation does the opposite.

MindWorX wants to translate then scale because he wants the scale to apply to the translation. But that's what Translation * scale does for me

With this setup, it should do it, yes.

oh wait, perhaps your W-coordinate is 0

it needs to be 1 for the translation to get applied

Shouldn't it be 1 by default for newly created vectors?

... and renormalised whenever it isn't 1 after an operation.

I don't know. In OpenTK, new Vector4(new Vector2(5,5)) = (5, 5, 0, 0)

I guess it depends on the implementation

I guess it doesn't bother with normalisation for 2D vectors. :D

... or maybe it just sets w=0 => vector is 2D.

I imagine it just defaults everything to zero because that's what people usually expect

because a Vector4 isn't necessarily always a homogeneous coordinate either.

@Jimmy Maybe it just differentiates between a vector (which can be rotated and scaled, but not meaningfully translated) and a point. I wouldn't know, didn't look much into OpenTK yet.

I love LINQ

var ray =
        this.world.ActiveRealm.Raytrace(new Vector2(this.world.ActiveRealm.RenderCenter.X, this.world.ActiveRealm.RenderCenter.Y), new Vector2(inputPoint.X, inputPoint.Y))
        .Select(point => new Point(point.X / TileWorld.TILE_SIZE * TileWorld.TILE_SIZE, point.Y / TileWorld.TILE_SIZE * TileWorld.TILE_SIZE))
        .Distinct()
        .ToList();

First I perform a raytrace, convert the coords to tile cells, remove the duplicates and return it as a list.

I guess I don't need to convert it to a list though.

nice

What does Select() do?

Ah, it's Map. Just found it.

Silly Microsoft using non-standard names for things.

Microsoft tried to make the names look like Sql

so map is select, filter is where

reduce is aggregate