nds - iphone - homebrew

starting with android

2011-12-29T12:23:00.000-08:00

After an Iphone and the Appstore, in between a Sony, now an android because of the IOIO microcontroller.

starting with Eclipse, the first android tutorial:

http://www.higherpass.com/Android/Tutorials/Writing-A-Basic-Android-Application/

and more:

http://www.higherpass.com/Android/tutorials/

but doing buttons and relative layout is not really JAVA....
for JAVA in android this link:

http://mobile.tutsplus.com/tutorials/android/java-tutorial/

After that, usually, one wants to bounce a ball:
http://www3.ntu.edu.sg/home/ehchua/programming/android/Android_2D.html
using a View with a ball inside, and showing extra possibilities.
After adding touch, all is wrapped up in an OOP fashion.

Some basic templates for Iphone coding

2010-05-29T23:58:00.000-07:00

Starting up a simple Iphone app, without much coding...:-)

Step 0:
a link to a photoshop template:
http://www.teehanlax.com/blog/2009/06/18/iphone-gui-psd-30/

Step 1:
download the 3GB Xcode with simulator:
http://developer.apple.com/technology/xcode.html

Step 2:
A very basic template with 5 backgrounds is downloadable here:
ExampleTemplateView5.zip

Step 3:
Unzip the folder and look at the contents:

and all the possible other steps:-)

start with editing the pictures, 320 x 480 pixels, and save the changed pictures in the folder under the same names,

OR, change the names in the folder AND in the code, see next image:

You can also add images immediately to the xcode. Then these images have to be dragged in the space under the other images in the space at the left side of the xcode window, so under Groups and Files and Resources...see this image:

After dragging the image in the right place, you get a window asking to add the image to the destination folder: check Copy radiobutton...

Then get the simulation running by clicking the green button with the hammer, the BUILD

When changing the images do a STOP and then again a BUILD.
Other example with the splash, the middle images and the finishing page:

download a template with splash, (starting screen) some basic backgrounds, (in a slideshow) and a finishing page:

ExampleTemplateView4.zip

the bit of code you have to add to get your image on top of a background:

UIImageView *imageAdded = [[UIImageView alloc] initWithImage: [UIImage imageNamed:@"myImageName.jpg"]];
[background addSubview: imageAdded];
imageAdded.center = CGPointMake(160,150); //position of center
imageAdded.alpha = .7;//transparency

-----

the backgrounds can be found in the file GameOverViewController.m

under " - (void)touchesBegan:"

more examples and templates on www.iphoneinsandouts.wordpress.com

on this last blog you can find more templates, tricks and design links. We did a class of about ten weeks and we have two app's in the appstore as a result of this. All kind of examples can be found with source code!

Check out two results: Google on "Aaipet appstore", and "Chameleon feeder appstore"

let us all make FREE games for ALL!!!!!!!

Iphone programming basics

2010-02-17T13:11:00.000-08:00

Starting with Iphone development you have to get used to some conventions in Objective C.

First you need some output for debugging in the Debugger Console, this is done with NSLog:

( By the way, A string is preceeded by @ )

NSLog(@"read the name");//simple string output in Debugger window.
Where to find this window: here, when you are running.....: (the gdb black window button...)

Now you can see what your code is doing:

NSLog(@"look at this number %d", intNumber); //int
NSLog(@"touched loc %f %f", currentTouch.x , currentTouch.y ); //float

showing a string in NSLog:
done by using stringByAppendingString:
NSLog([@"see how the name of the player is changed in setNamePlayer :" stringByAppendingString:@"new name of player"]);

mind the function clammers: [ and ] ..... [ ... stringByAppendingString: ... ], this gives a String, because this comes from the class NSString.

simple numbers: (no pointers needed)
int, float, bool
int myInt = 9;
float myFloat = 9.123;
bool myBool = false;

Declaration: int (float, bool)
in the header:

@interface ScoreObject : NSObject {
int playingFieldNumber;
}
@property int playingFieldNumber;

in the source:
#import "ScoreObject.h"
@implementation ScoreObject
@synthesize playingFieldNumber;

Number Objects: (always pointers in Objective C)
making an Number Object (for instance needed populating NSArray or NSMutableArray
NSNumber* myNumberObject = [NSNumber numberWithInt: 99 ];
NSNumber* myNumberObject = [NSNumber numberWithFloat: 99.99 ];

getting the value back from an NSNumber Object:
int myIntValue = (int)[myNumberObject intValue];
float myFloatValue = (float)[myNumberObject floatValue];
bool myBoolValue = (bool)[myNumberObject boolValue];

using this in NSLog:
NSLog([@"look at this number %d", (int)[myNumberObject intValue]);

Declaration: objects like NSNumber, NSArray
in the header
@interface ScoreObject : NSObject {
NSMutableArray *fieldNameArray;
}
@property (nonatomic, retain) NSMutableArray *fieldNameArray;

in the source:
@implementation ScoreObject
@synthesize fieldNameArray;

Last thing about NSLog... how to get an object out?

NSLog( @"this is my object: %@", myNSNumber );

of course you do not get a int float bool, like this, but an object name.]]>

Provisioning profile blues

2010-02-06T23:00:00.000-08:00

I was cooking up exercises now with the provisioning profiles. Just take a new clean view based project, nothing in it but the grey screen and see where you get with the different provisioning profiles, and debug release and distribution on your device.

Later on I saw someone else already having a full diner of these exercises!
http://www.codza.com/how-to-fix-iphone-code-signing-errors

Why there are two Info windows, one for the project and the other for the target, why you have to insert again a special Entitlements.plist, why you have this possibility of debugging in the Entitlements but have to uncheck it.....stopping the app running from Xcode on your device....

I hope sometimes the real apple enlightenment will shine on me...

Iphone Provisioning Profile problems links

2010-02-03T00:55:00.001-08:00

While trying to submit my app I had to search for links explaining all the details of the weird security system of the Iphone apps, in combination with Xcode.
It really took me days and this signing is much more difficult than programming to get it right.
(Studying relativity theory is easier than getting this right, and a lot more fun!)

This is a collection of links I found useful:

http://www.24100.net/2009/02/iphone-sdk-mobile-provisioning-0xe800003a-0xe8000001/

http://www.iphonedevsdk.com/forum/iphone-sdk-development/5429-codesign-verification-nightmare.html

http://johnehartzog.com/2009/04/iphone-app-ad-hoc-gotchas/

The first blog I found with an Entitlements.pList written out is:

http://adrianhosey.blogspot.com/2009/01/iphone-development-provisioning.html

In this blog you get a picture of the contents of this file, only for me:
debugging has to be allowed!!! (in debugging) Other tell to uncheck this.(for release)

In release (device) the app is loaded to my device, but not started, I have to start it myself.

Then I discovered there were two files with seperate properties: (called Info)

double click on Project to get the Project Properties: Project Info

double click on the app icon under Target to get the Target properties: Target Info

These files are practically the same, but the Code Signing Entitlements is not taken over!!!

So for me one was at still at dist.plist the other on Entitlements.plist, giving an error!!!!

So we should set SIX files the right way.... (debug release distribution, for simulator and for device.....)
Who has cooked this system up?

It must have been the Tetris God, messing up!
(see him here: http://www.youtube.com/watch?v=Alw5hs0chj0)

Debugging a double free malloc error

2009-12-24T22:36:00.000-08:00

When making an instance you have to free it again with release. But there is also for some familiar things like CGArray an auto release.
I had an error in the Debugging Window telling me there was a double release....
malloc: *** error for object 0x1346f90: double free
*** set a breakpoint in malloc_error_break to debug
where to put what breakpoint?
in gdb? well somehow this didn't work for me.

The problem is of course the object is released later on, not stopping at the line which caused the problem....

these links helped me setting the environmental variables needed to debug:
http://lists.apple.com/archives/Xcode-users/2008/Apr/msg00160.html
http://www.cocoadev.com/index.pl?DebuggingAutorelease
http://www.cocoadev.com/index.pl?MallocStackLogging
http://www.cocoadev.com/index.pl?NSZombieEnabled
http://www.cocoadev.com/index.pl?DebuggingTechniques

after setting the environmental variables given in the link I got this error:
*** -[CFArray release]: message sent to deallocated instance 0x135e0a0
which didn't tell me anything new, the instance number I could not trace...
the debugger told me this:
malloc: stack logs being written into /tmp/stack-logs.373.Grungegolf.KT6xux
but where is this file situated?

this command didn't help because the pin number (in this example 3939) from this example is not in my gdb error

shell malloc_history 3939 0xa4e10

this didn't help either:
gdb: t a a bt

in the end I searched for all the release commands I have given with the only clue that is must be an Array, and I found the culprit...

trying to understand:
I loaded a textfile:
NSString *myText = [NSString stringWithContentsOfFile:whateverFilePathString];
I used this to load my game fields, being all in the textFile, a String
so seperate the fields:
NSArray *allData = [myText componentsSeparatedByString:@"***"];
with the separator "***"
then taking element 3 which contained the bouncers...
NSArray *bouncerData = [[allData objectAtIndex:2] componentsSeparatedByString:@","];
putting the bouncer data, making them numbers in another array of objects...
I released the bouncerData (wrong!)
I suppose this bouncerData array is released with the allData array, which is also released automatically....

NSArray features

2009-10-27T01:02:00.000-07:00

We had an NSArray of objects of the type UIImageView.

the array was retained being a property etc

initialize like this:

testBall2Array = [NSMutableArray arrayWithObjects:
[[UIImageView alloc] initWithImage: [UIImage imageNamed:@"testBall2.png"]],
[[UIImageView alloc] initWithImage: [UIImage imageNamed:@"testBall2.png"]],
[[UIImageView alloc] initWithImage: [UIImage imageNamed:@"testBall2.png"]],
[[UIImageView alloc] initWithImage: [UIImage imageNamed:@"testBall2.png"]],
[[UIImageView alloc] initWithImage: [UIImage imageNamed:@"testBall2.png"]],
nil];

We succeeded in putting the views on screen in the same method but apparently in another method we got a bad access error when doing something with an element of this array.

The elements were not retained, but how to retain them?

This is one of the niceties of Xcode! :-(((

The answer is found here:
http://stackoverflow.com/questions/519467/iphone-getting-the-uiimageview-back-from-nsarray

you have to initialize with a different method:

initWithObjects

instead of

arrayWithObjects

the first retains the objects the second doesn't......

well this is the way you spent your time doing Xcode....

NSMutableArray and CGPointArray

2009-10-24T04:58:00.000-07:00

We wanted to set up an array of CGPoints but failed, in the properties....
We had to do it with a NSMutableArray. But how to get the CGPoints in?

the answer is found here:

http://stackoverflow.com/questions/943945/how-to-form-cgpoint-array-in-objective-c

so:

@interface OrbitLinesDrawView : UIView {

UIColor *strokeColor;
UIColor *rectColor;
CGFloat strokeWidth;
CGFloat cornerRadius;

NSMutableArray* orbitLinePointsArray ;
}

@property (nonatomic, retain) NSMutableArray* orbitLinePointsArray ;

and then in the .m file:

@synthesize orbitLinePointsArray;

in the init:

orbitLinePointsArray = [[NSMutableArray alloc]init ];

for (int i = 0; ... etc....)
CGPoint aPoint = CGPointMake(100 - 10*i, 200+10*i);
NSValue* pointValueObject = [NSValue valueWithCGPoint:aPoint];
[orbitLinePointsArray addObject:(NSValue*)pointValueObject];
}

(dont forget to alloc the orbitLinePointsArray)

for setting the CGPoints later on:

-(void)setObjectInPointArray: (int)positionInArray : (CGPoint) myPoint
{
NSValue* pointValueObject = [NSValue valueWithCGPoint:myPoint];
[ orbitLinePointsArray replaceObjectAtIndex:positionInArray withObject:pointValueObject ];
}

drawing the points in the

-(void)drawLines {

...
...
...

CGPoint lineArray[2];

for(int i=0 ; ... etc....)
{
lineArray[0] = ((CGPoint) [[orbitLinePointsArray objectAtIndex:i] CGPointValue]);
lineArray[1] = ((CGPoint) [[orbitLinePointsArray objectAtIndex:i+1] CGPointValue]);

CGContextAddLines ( context, lineArray, 2 );
CGContextStrokePath(context);
}

[self setNeedsDisplay];
}

- (void)drawRect:(CGRect)rect {

[ self drawLines ];

}

so Objective C, Cocoa, ....whatever....wrap it up in further and further objects, that seems to be Apple.....

beta testing on the iphone

2009-08-21T22:05:00.000-07:00

a list of probably useful links for setting up a beta test of a new app on the iphone:

the most important word is "Ad Hoc"

maybe the best is just to follow the list in the developer.apple.com -> program portal -> Distribution tab (in the link of team, certificates, devices, app id's , provisioning)
the problem is that I at least tend to search the web for links about this subject before looking in the apple dev portal....

in distribution tab open up the prepare tab and read the steps...

problem with me was the key - distribution profile combination
in xcode in projectpreferences the distribution profile was indicated but also the error of no combination of key and profile found
in the keychain i had to select a private key and drag the profile into it, than the key was combined with the profile like on the image:

other not apple link info:

intro:
http://www.appletell.com/apple/comment/app-store-development-part-2-provisioning-and-ad-hoc/
http://www.appletell.com/apple/comment/iphone-developer-intervention-week-7-beta-testing/

how to make an ad hoc provisioning profile:
http://www.manyfriends.com/wiki/iphone/index.php/Create_an_ad-hoc_provisioning_profile#Create_the_ad-hoc_profile

a list of steps to follow after that
http://furbo.org/2008/08/06/beta-testing-on-iphone-20/

a free app to use to get betatester to send you the UDID
http://phobos.apple.com/WebObjects/MZStore.woa/wa/viewSoftware?id=285691333&mt=8

there is a site for global beta testing (not tested by us yet:-)
http://ibetatest.com/iphone/controllers/welcome/

Sound on the Iphone

2009-06-22T22:42:00.000-07:00

Having lot's of trouble gettting a good sound for my game on the iPhone, due to the special sound format. The Iphone explanation didn't help much, the code is clear, no it was the formatting.
I got a sound ending in a terrible noise, when played on the Iphone.

Finally i found the right links, explaining and suggesting the solution:
http://www.wooji-juice.com/blog/iphone-openal-ima4-adpcm.html

here something is told about the IMA4 format, and alos how to get it using the terminal window:

afconvert -f caff -d ima4 audiofile.wav

usinf this command i got a sound working normally in my game.

You have to know how to use the terminal in Mac OS (personally i am a Windows user :-)

here is a link explaining how to get in the right directory:

http://macapper.com/2007/03/08/the-terminal-an-introduction/

this amounts to:
opening the terminal using spotlight...
cd Desk*
cd FileWithSoundOnDeskTop
and then the command above afconvert

So in the end, all was already on my MACBOOK (with the double booth) only I had to find it...

have fun!

C special on arrays and functions

2009-04-21T17:01:00.000-07:00

Introductions of C can be found on the internet. These explain basic stuff like pointrs and arrays. But programming a bit for the nds, you need details which are not in the introductions.

Programming in Programmers Notepad with testing in an emulator (all for the nds in C) is not the easiest way of experimenting. So sometimes i write some test code in Dev Cpp.

This is an example: for one of the field in a minigolf game you need an array. This array must be initiated in a struct. This struct has a function making this initialization. Many possibilities of doing this, of course. One of the possibilitites is shown here:

This is (part of ) the struct: (in a header file)

typedef struct Field{

int pointsList[200];
int roundNr;

}FIELD;

extern FIELD makeField( FIELD myField, int p[ ]);
extern void printField(FIELD myField);

The function is found in the source file: see especially how the array is called. p[ ] is a pointer.

FIELD makeField(FIELD myField, int p[ ], pSize)
{
myField.roundNr = 4;//just some value for this example

int c1;
for ( c1=0; c1 < pSize ; c1++ )
{
myField.pointsList[c1] = (int)p[c1] ;
printf("value from field make function %d \n " ,(int)p[c1] );
}
return myField; //necessary to get the object back, this is C
}

void printField(FIELD myField)
{
int c1;
for ( c1=0; c1 <>
printf("from print %d \n",(int)myField.pointsList[c1] ) ;
}

then the main file:

#define NUM_ELEM(x) (sizeof (x) / sizeof (*(x))) //this trick is explained see below

int main(int argc, char *argv[])
{
FIELD myField;
int p[ ] = {1,2,3,4,5,6,7,8,9,0};
myField = makeField( myField, p, NUM_ELEM(p)); //return because of passing by value
printField( myField);//check
}

the NUM_ELEM is explained in http://en.wikibooks.org/wiki/C_Programming/Pointers_and_arrays#Pointers_in_Function_Arguments
nearly at the end of this page.

You see how the array is passed to the function, how the struct object is filled with the array. All kinds of little C details dealing with the not OO world!

Of course you can do this in another way, not returning the field, you have to use a pointer to this struct:

in the main:

FIELD * myFieldPointer;
myFieldPointer = &secondField;
int p2[] = {11,12,13,14,15,16,17,18,19,20};
makeFieldPointer( myFieldPointer, p2, NUM_ELEM(p2) ); //not returning because of passing by reference
printField( secondField);//check

You need another function makeFieldPointer:

void makeFieldPointer( FIELD *myFieldPointer, int p[ ], int sizeP)
{
myFieldPointer->roundNr = 4; //need the -> because of the pointer
int c1;
for ( c1=0; c1 <>
{
myFieldPointer->pointsList[c1] = (int)p[c1] ;
printf("value from makeFieldPointer %d \n " ,(int)p[c1] );
}
}

Latest DevKitPro

2009-04-21T16:55:00.001-07:00

Devkitpro is updated. This means getting used to some changes, in naming, some new functions, some new possibilities. Not always what you are waiting for! But ok, here some pages which make life easier:

Tis is a page with a usefull list of registers:
http://www.coranac.com/2008/09/11/nds-register-overview/

then this page under point 10 explains other changes:
http://quirkygba.blogspot.com/2008/12/upgrading-to-devkitarm-r24-on-linux.html

if the registers, for instance of backgrounds are changed, and the irqinit() is removed, most of the stuff works as before. I had more difficulties getting the maxmod sounds work with the DSerial, since both of this is going through the ARM7 apparently.....my solution was statistics, but that is not very neat....

Using libfat

2009-04-12T07:33:00.000-07:00

The example in de devkitpro prints the directory of the card.
It only runs on the ds itself, not in the emulator i use....:-(

ok this is a diary, just stating the things needed and where i found these: i needed to read in the integers describing the points of the triangulation for the minigolf game fields.

the libfat example only described reading the directory.

I found an example of really reading in a textfile here: (actually it is written for the wii, but who cares?)
http://www.codemii.com/2009/03/02/tutorial-10-using-the-filesystem/

and the functions to read a txt file are: (adding to the example in the devkitpro)

void die(char *msg) {
perror(msg);
fatUnmount(0);
exit(0);
}

void readFile()
{
FILE *f = fopen ("hello.txt", "rb");// If file doesn't exist or can't open it then we can grab the latest file
if (f == NULL) {
die("Could not open myfile.txt file for reading.\n");
}

else {
char file_line [20];
int line_number = 0;
while (fgets (file_line, 20, f)) {
printf("Line %i: %s\n",line_number, file_line);
line_number++;
}
fclose(f);
}
}

as explained (and needed for my minigolf, reading in the triangulations of the fields)
you can easily read in and distribute integers to arrays lik this:

make these arrays global, for instance

int x_pos[] = { 0, 0, 0, 0, 0};
int y_pos[] = { 0, 0, 0, 0, 0};

void readFile()
{
FILE *f = fopen ("hello.txt", "rb");
// If file doesn't exist or can't open it then we can grab the latest file
if (f == NULL) {
die("Could not open myfile.txt file for reading.\n");
}
else {
char file_line [20];
int line_number = 0;
while (fgets (file_line, 20, f)) {
printf("Line %i: %s\n",line_number, file_line);

char *temp_string;
temp_string = strtok (file_line, " ");
if (temp_string != NULL) {
x_pos[line_number] = atoi(temp_string);
temp_string = strtok (NULL, " ");
if (temp_string != NULL) {
y_pos[line_number] = atoi(temp_string);
}
}
line_number++;
}
fclose(f);
}
}

which still prints the values as a check, this can be left out.

Writing to a file is also described above, which is useful for a highscore, eventually.

Writing in the example went well, but for the fact that the integers were written without a space, so the lines glued together.

This was solved adding:
char end_string[2] = { 0x0D, 0x0A};
fputs (end_string ,f);

so the function becomes:

void writeFile()
{

FILE *f = fopen ("myScorefile.txt", "wb");

int line_number = 5;

if (f == NULL) {
die("Could not open myScorefile.txt file for writing.\n");
}
else {
int x;
for (x = 0; x < line_number; x++) {
char temp_string[20];
sprintf(temp_string, "%i %i", x_pos[x], y_pos[x]);
fputs (temp_string ,f);
char end_string[2] = { 0x0D, 0x0A};
fputs (end_string ,f);
}
fclose(f);
}
}

DSerial Edge

2009-04-01T12:11:00.000-07:00

DSerial Edge is something very interesting. It can make the nds explore the outer world!

documentation is found here: http://www.natrium42.com/wiki/DSerial_Edge

we did a first project linking switches made in or from books with an app on the nds

the only problem is at the moment:
dserial libs and maxmod sound libs interfere somewhere, so working with dserial limits the nds possibilities seriously!

On a small testapp, we connected the ports of IO1 to HIGH. This works fine, when a low is presented a low is seen:

dsePinWrite(0, 6, true);
dsePinWrite(0, 7, true);
dsePinWrite(1, 1, true);
dsePinWrite(1, 0, true);
dsePinWrite(1, 2, true);
dsePinWrite(1, 4, true);
dsePinWrite(1, 5, true);
dsePinWrite(1, 6, true);
dsePinWrite(1, 7, true);

dsePinMode(0, 6, INPUT);
dsePinMode(0, 7, INPUT);
dsePinMode(1, 1, INPUT);
dsePinMode(1, 0, INPUT);
dsePinMode(1, 2, INPUT);
dsePinMode(1, 3, INPUT);
dsePinMode(1, 4, INPUT);
dsePinMode(1, 5, INPUT);
dsePinMode(1, 6, INPUT);
dsePinMode(1, 7, INPUT);

When playing sounds, the ports are steered LOW by the sounds...using the same ARM7 ????

this goes very fast, iif this is played in the loop:

if( dsePinRead(0, 6)) iprintf("\x1b[6;0H 06=HIGH"); else iprintf("\x1b[6;0H 06=LOW");
if( dsePinRead(0, 7)) iprintf("\x1b[7;0H 07=HIGH"); else iprintf("\x1b[7;0H 07=LOW");

if( dsePinRead(1, 0)) iprintf("\x1b[9;0H 10=HIGH"); else iprintf("\x1b[9;0H 10=LOW");
if( dsePinRead(1, 1)) iprintf("\x1b[10;0H 11=HIGH"); else iprintf("\x1b[10;0H 11=LOW");

if( dsePinRead(1, 2)) iprintf("\x1b[11;0H 12=HIGH"); else iprintf("\x1b[11;0H 12=LOW");
if( dsePinRead(1, 3)) iprintf("\x1b[12;0H 13=HIGH"); else iprintf("\x1b[12;0H 13=LOW");

if( dsePinRead(1, 4)) iprintf("\x1b[13;0H 14=HIGH"); else iprintf("\x1b[13;0H 14=LOW");
if( dsePinRead(1, 5)) iprintf("\x1b[14;0H 15=HIGH"); else iprintf("\x1b[14;0H 15=LOW");

if( dsePinRead(1, 6)) iprintf("\x1b[16;0H 16=HIGH"); else iprintf("\x1b[16;0H 16=LOW");
if( dsePinRead(1, 7)) iprintf("\x1b[17;0H 17=HIGH"); else iprintf("\x1b[17;0H 17=LOW");

then you barely see it, but placing this:

if( !dsePinRead(0, 6)) iprintf("\x1b[6;0H 06=LOW");
if( !dsePinRead(0, 7)) iprintf("\x1b[7;0H 07=LOW");

if( !dsePinRead(1, 0)) iprintf("\x1b[9;0H 10=LOW");
if( !dsePinRead(1, 1)) iprintf("\x1b[10;0H 11=LOW");

if( !dsePinRead(1, 2)) iprintf("\x1b[11;0H 12=LOW");
if( !dsePinRead(1, 3)) iprintf("\x1b[12;0H 13=LOW");

if( !dsePinRead(1, 4)) iprintf("\x1b[13;0H 14=LOW");
if( !dsePinRead(1, 5)) iprintf("\x1b[14;0H 15=LOW");

if( !dsePinRead(1, 6)) iprintf("\x1b[16;0H 16=LOW");
if( !dsePinRead(1, 7)) iprintf("\x1b[17;0H 17=LOW");

all go low immediately. This speed can make a work around, only triggering on longer intervals of LOW?

anyone a more fundamental solution?

workshop C II

2009-02-27T03:55:00.000-08:00

In the first workshop we saw that C was about the same as JAVA, although C is not object orientated, in this workshop we see that there are a few differences...
To start with a real hurdle coming form JAVA we look at pointers.

pointers
http://home.netcom.com/~tjensen/ptr/pointers.htm

a pointer variable refers to an address in memory

int k; //makes an integer, not setting the value yet.

int *ptr;//makes an address variable, not setting this address either
//set aside however many bytes is required to store an address in memory

ptr = &k; //give the address variable the value of the address of the integer

*ptr = 7; //* is here the dereferencing operator
k = 7;//this last two lines have the same effect

pointer arithmetic is based on the size of the pointer's type
ptr++;//refers to the next address in memory, but shifts the space of an integer.

short m;
short *ptrShort = 6;
ptrShort++;//this would shift the space of a short

C will not permit the comparison of a pointer to type integer with a pointer to type character
that is why there are "void pointers"
void *vptr;// a generic pointer

sizeof(int) is a function which returns the number of bytes used by the type provided. This is typically used in pointer arithmetic.

arrays

Arrays were introduced in workshop 1. But there is more than meets the eye, because arrays are big structures and for these structures C uses pointers.
So arrays and pointers are very related.

this is still like JAVA:
int my_array[] = {1,23,17,4,-5,100}; //integers stored continuously

But now we see the difference:
int *ptr;
ptr = &my_array[0]; /* point our pointer at the first integer in our array */

int i = 2;
my_array[i] refers to the same variable as *(ptr + i)) : 17

note: an arrayname is a pointer, although it seems a "normal: variable:
&arrayName[0] can be replaced with arrayName

then there is a curious C effect:
char myArray[5] = {'h', 'o', 'm', 'e', 'b', 'r', 'e', 'w', '\0'};
myArray[3] = *(myArray + 3) = 'm';
3[myArray] = 'm'; // because *(i + a ) = *( a + i) !

remember the multidimensional array? This is a pointer thing too, of course:
char multi[5][10];
*(*(multi + row) + col) is the same as multi[row][col]
because the elements are stored continuously in memory

to work with "raw" pointers, in pointerarithmetic and wanting to know exactly the place of a variable you use sizeof(int).

structs

http://computer.howstuffworks.com/c31.htm
http://www.cprogramming.com/tutorial/lesson7.html

Although there are no objects, we can make or own complex variables, with members:
You can write your own data structure!

example:
typedef struct Triangle{

POINT p[3]; //member in this case another struct,
LINE l[3];//another struct as a member

int t[5];
int neighbours[25];
int outsidePoint[25];
int bouncers[25];

int neighbourNumber;
int outsideNumber;
int bouncerNumber;

}TRIANGLE; //this gives the struct a special name.

//so POINT en LINE zijn andere structs (convention = capitalized name)

struct Triangle myTriangle;//declaration or
TRIANGLE myTriangle;

members are defined with a point:
myTriangle.bouncerNumber = 7;

if you have a function you pass not the struct but a pointer to the struct:

*TRIANGLE myTrianglePointer;
myTrianglePointer = &myTriangle;

defining a member would become:
(*myTrianglePointer).bouncerNumber = 63;//but this can be written as:
myTrianglePointer-> bouncerNumber = 63;

strings/characters

a string is an array of characters terminated by '\0'

char my_string[40];

my_string[0] = 'T';
my_string[1] = 'e';
my_string[2] = 'd':
my_string[3] = '\0';

char my_string[40] = {'T', 'e', 'd', '\0',};

char my_string[40] = "Ted";

all these lines do the same thing, mind the different quotes.

standard string functions:
strcopy();
strlen();
strcat();
strchr();

functions
We already know of the pointer, and why we use pointers. In functions this is used:
Never send the struct itself, but always a pointer to the struct.
Of course this means that the struct itself will be changed in the function.

This is the difference:
passing by value
passing by reference -> pointers
since an arrayName is a pointer, fie(arrayName) passes the array by reference.

header files
You can write your whole project in the main, using functions on top (above) the main.
This can become rather a mess.
In C you can order your project in files seperated from the main.m

For instance, you can have a game.m file.
Then you have a header file for initialization of the functions used in the game.m file: game.h

the header files are collected in the folder called include, and the .m files in a folder called source.

You have to include the header files to let other files see the functions.

A problem can be that a header file is included several times, therefore in C you use a special header definition:

ifndef
#define
...
#endif

this prevents including the header twice, which would cause an error on function twice defined.

#ifndef __BACKGROUND_H__
#define __BACKGROUND_H__

#include 

extern void initBackgrounds();
extern void fadeBackground(short screen, bool hide);
extern void setBitmapBackground(short screen, const void *bitmapData, int bitmapSize);
extern void setTiledBackground(short screen, const void *tileData, int tileSize, const void *mapData, int mapSize, const void *paletteData, int paletteSize);

#endif

Introducing nds: workshop 1

2009-01-23T07:22:00.000-08:00

Workshop 1: (draft version)

1. What is nds?
2. Installing the emulator and playing nds programs
3. Installing IDE, and other development tools (DevkitPro, Grit, Sox)
4. Basic C (variabels, types, iterations, selections, array's, includes)
5. Make your first program (what about....Hello World?)

1. discovering the console nds

it has two screens!
it has so many buttons
it has a touch screen
it has a 3D mode
it has two processors! But these two processors are not the same.
you can program it yourself!
how to get a program on the nds

2.

Nocash GBA Emulator: NO$GBA.EXE
*most reliable
NDeSmuME
*first notions discovering the memory: viewing maps, registers, (dont understand this now :-)

3.
Installing the DevkitPro, working with Programmer's NotePad, Wingrit, grit and Sox.

4. A first bite of C. Is it like JAVA?

variables:
The memory of the nds is limited, so try using as small variables as possible

(even smaller variables will be introduced later :-)

char, of type character 8 bits
short, int, long, of type integer 16/32 bits
double, float of type real 32/64 bits

then you have signed and unsigned:

short -32,768 to 32,767
unsigned short 0 to 65,535, the thing is simply shifted

declarations: you have to declare variables before using them

int start, time1, time2;

special numbers can be defined like this:

#define PI 3.141592654

Arrays (actually pointers, but about pointers later)

int myarray[10];
unsigned int list[5] = { 10, 15, 12, 19, 23 }; //giving the elements right away!
float rdata[128], grid[5][5];

the first element is myArray[0];

asking an element of an array: a = myArray[3];

there is no boundery checking, because an array is a pointer to a position in memory.
What a pointer is: later

when you declare an array the elements are not initialized.

a loop with an array:

for( i = 0; i <= 127; i = i + 1 )
{
printf ( "\f\n", rdata[i] );
}

what is wrong? Yep!
you forgot this:
int i;

dynamically declaring an array:

int *myArray; // don't know exactly how big it will be....
//do a lot of stuff here....
myArray = malloc(10*sizeof(int));

be careful to free the memory again:
free( myArray ); //when aloc then free

while loop:

while ( i <= 100 ) /* loop until i > 100 */
{
printf ( "the number %i \n ", i );
i++;
}

structure of a well known C program:

#include <>
void main()
{
printf("\nHello World\n");
}

conditionals:

if(i <>88)
{
i=88;
}

if ( i ==88)
printf("\n88\n");

the switch is sometimes handy:

switch (i)
{
case 1:
{
printf("it was a one!");
break;
}
case 2:
{
printf("it was a two");
break;
}
default:
{
...block of statements..
}
}

5.
look closer at small example programs in the DevkitPro example folder, running on the emulator, practice some C, using the printing of characters on the upper or lower screen, switching between the screens:

C programming: make the order like this:

void setup()
{
//prepare your stuff
}

void loop()
{
//do something a lot of times
}

main()
{
setup();
loop();
}

The problem with every programming is not only the language but also the special effects.....special methods written for this language to serve the purpose faster...more comfortable sometimes, but these methods have to be learnt too....like for instance consoleInit

assignment 1: now restructure the HelloWorld example in the examples of DevkitPro like this setup-loop scheme.

assignment 2:make the printBothscreens like this setup-loop scheme (this will be more difficult because the consoleInit is rather specialized, be careful!)

have fun!

Xcode Iphone Quickies

2009-01-17T04:04:00.001-08:00

found some quickies for meself :-)

making an int from NSNumber:
int myInt = (int)[(NSNumber *)numberObject intValue];
also floatValue

making NSNumber from int:
NSNumber *mynumberObj = [NSNumber numberWithInt: 255]

making a String from int:
NSString *myString = [NSString stringWithFormat:@"%d", myInt];

making an int from a String:
int myInt = [mySTring intValue];

concatenating two strings:
NSString *myString1 = [@"field" stringByAppendingString: myString1];

than a useful blog showing how to draw on a view in all details, lines, text, an image
http://trailsinthesand.com/exploring-iphone-graphics-part-1/
(you have to mirror the text, for instance....text is drawn in reverse in default.)

while drawing like an artist you might run out of space, then you need the UIScrollView, this is a bit .... peculiar too: it seems you have to trick the interfacebuilder in believing the wrong things: first connecting the outlets, then changing the File's Owner to UISCrollView and it shows "errors" with the former outlets, while they work fine (?) Anyway here is a good tuto:
http://blog.atrexis.com/index.cfm/2008/9/29/Interface-Builder-Tutorial

minigolf on the iphone

2009-01-13T22:37:00.000-08:00

We started a minigolf on the nds, learning a lot about the way C uses arrays. The same process started with Xcode-Objective C on the Iphone, recreating this minigolf game.

The minigolf has a small engine allowing for shapes of the field which are queer, not rectangular.
For this we needed to triangulate the field and have arrays ready of the points, the lines and the triangles involved. The triangle class has an pointarray inside, as well as the three lines in an array.

In the nds code, then the arrays were initialized like:

u16 p3[] = { 61,180, 22,171, 8,102, 8,42, 55,7, 95,13, 105,34, 136,69, 156,84, 164,83, 184,79, 188,74, 183,56, 179,52, 157,45, 151,45, 117,82, 114,87, 119,111, 120,115, 162,147, 160,166, 133,180, 84,144, 76,109, 76,79, 131,9, 196,6, 241,46, 237,74, 220,102, 198,125, 163,133, 127,103, 120,88, 84,53, 77,54, 64,66, 61,74, 60,132, 63,138, 86,166, 0,0,0,0,0};

each three integers were a triangle, the naughts at the end were to stop the making triangle objects, C having no way of telling how long an array is (?).

of course these data should be read in from an external file, so giving the possibility to add fields of your own. In Xcode i decided to do that immediately:

reading in a .txt file: (The triangles are given only with the number of the points from a point array, so 1,2,3 means triangle with point number 1, point number 2, point number 3.)

NSString *filePath2 = [[NSBundle mainBundle] pathForResource:@"triangles1" ofType:@"txt"];
if (filePath2) {
NSString *myText = [NSString stringWithContentsOfFile:filePath2];
if (myText) {
NSLog(myText); //see if it worked!
NSArray *arr = [myText componentsSeparatedByString:@","];

and at the end separating the file with the comma's in between the numbers.
Then the points were searched from the point array made earlier.

ok before this code: don't forget to inilialize the array:
triangleArray = [[NSMutableArray alloc] init];

now use a loop to generate the set of three points:
integer_t i;
for(i=0;i
{
int triangleNumber1 = (int)[[arr objectAtIndex:i] intValue ];
int triangleNumber2 = (int)[[arr objectAtIndex:i+1] intValue ];
int triangleNumber3 = (int)[[arr objectAtIndex:i+2] intValue ];

make the triangle object: (using the point array made earlier in the same way)

Triangle *myTriangle = [[Triangle alloc] initWithNumbers3:( MyPoint *)[pointArray objectAtIndex:triangleNumber1]:( MyPoint *)[pointArray objectAtIndex:triangleNumber2]:( MyPoint *)[pointArray objectAtIndex:triangleNumber3] ];

and adding this triangle object to the triangleArray:

[triangleArray addObject:myTriangle ];
}
NSLog(@"made triangleArray");
}
}
else
NSLog(@"no path found");

of course without the actual point, line and triangles classes you cannot do much with this code, but is sketches the principle. (This blog is a diairy.....what is written in 5 minutes....could have costed me a few houres, mainly searching for the right syntax in a for me "new" language, in all kinds of blogs on the web.)

reference link array:
http://www.cocoadev.com/index.pl?NSMutableArray
reference link reading in an external file (scroll down):
http://iphoneincubator.com/blog/page/2
reference link for NSString "Quickies"
http://www.borkware.com/quickies/one?topic=NSString

Iphone and multiple (sub) views

2009-01-11T08:07:00.000-08:00

After quite a few tutorials and a first go at the minigolf game for the iphone (started at the nds - not yet finished :-) I asked myself if it was possible to combine more views. This could facilitate the game and is great for overlaying drawing.

At first i searched and found an example of a rounded corner view here:
http://iphonedevelopment.blogspot.com/2008/11/creating-transparent-uiviews-rounded.html
but i didn't succeed in implementing this in my proto minigolf.

Eventually i decided to try with a new clean project Viewbased.
I added three views in the folder classes
then i went to the interfacebuilder and took the ViewTestController.xib

I loaded this nib with 3 different views in the main view (see image) i then connected the type of the three subviews to my newly added classes, i colored the classes differently.
saved the file, returned to Xcode and compiled,
to my surprise this really simple procedure was accepted.
The roundedrect view showed up as predicted with rounded rects.
(In fact this is like overlaying ImageViews.)

My mistake was combining interfacebuilder with programmed views, and adding these views in the
/* // Implement loadView to create a view hierarchy programmatically, without using a nib. - (void)loadView {
} */
function, because it said to do so "to create a view hierarchy programmatically", which i apparently misunderstood.

I added in the rounded rect view a pentagon with this code:

CGContextRef context = UIGraphicsGetCurrentContext();
CGContextSetStrokeColorWithColor(context, [UIColor greenColor].CGColor);
CGContextSetLineWidth(context, 5.0);
CGPoint five[6];
integer_t i;
for(i=0;i<6;i++)
{
five[i].x = 100 + 30* sin( i*2*3.1415/5);
five[i].y = 100 + 30* cos( i*2*3.1415/5);
NSLog(@"myPoint 4 #%f #%f", five[i].x, five[i].y);
}
CGContextAddLines ( context, five, 6 );
CGContextStrokePath(context);

[When adding an UIImageView to one of the SubViews, the image when to big spills over the view. Just set the Clip Subview check to keep the image in bounds.

Could it be possible to make this too big image scrolling? Certainly! But since i have three views inside a view it is a bit more complicated than the example here:
http://www.vimeo.com/16421 50
(There must be a natural way. like the three views?) ]

from which i learned that the array CGPoint five[6] can be made without the need of a MSMutableArray, meant for NSObjects. It is like ordinary C here for once.

So slowly my C knowledge is becoming related to objective-C of Xcode

Iphone programming

2009-01-07T06:23:00.000-08:00

Or is this called scripting? :-)

I started with Xcode and Iphone programming quite naively thinking my C experience on the nds would help me out. Not at all! Xcode's Objective-C is quite different. Luckily there are a few tutorials around (they are everywhere - everyone seems to make about 10 a day). Following a tutorial is great, eventually it works, but you have no clue why. Idem for the interfacebuilder.
SO in the end only doing it yourself helps and then it becomes clear that the tutorials only give you the most simple situations.
Anyway some tuto's were helpful, here a few links to them:

The first step:
get to know the apple tutorials! These are the best found on the apple iphone dev site:
http://developer.apple.com/iphone/

lots of other tutorials can be found, for instance:
http://icodeblog.com/2008/10/20/iphone-programming-tutorial-using-uitouch-to-drag-an-image-around-the-screen/
http://www.bit-101.com/blog/?p=1784
http://www.iphonesdkarticles.com/2008/07/first-iphone-application.html
http://www.appsamuck.com
but i think, with hindsight that the apple examples are just the best.

following these tutorials is ok, but learning to work with the interface builder cannot be achieved this easy. For understanding the interface builder you could try these links:
http://en.wikipedia.org/wiki/Interface_Builder
http://developer.apple.com/tools/interfacebuilder.html

Then comes the moment that you start making your own classes. Objective C has a curious way to declare the variables, properties, and call functions with mutiple variables...
a few links can facilitate the process of getting used to Objective C:
http://cocoadevcentral.com/d/learn_objectivec/
http://www.otierney.net/objective-c.html
http://en.wikipedia.org/wiki/Objective-C

also arrays are used different from normal C, be careful!

The emulator of the iphone is very handy, because uploading to your iphone is quite an effeort with certificates and signatures (after paying the fee for becoming a developer)
Turning the emulator is easy with Command LEFT or RIGHT.

Uploading an application can be done, having the certificates and codes and signatures.
But jailbreaking or hacking is seems also possible. I tried but my MacBook is just the only MAC machine which refuses this apparently.
http://www.iclarified.com/entry/index.php?enid=1881
http://www.hackthatphone.com/114/iphone_step2_mac.html

LIBFAT

2008-12-20T01:41:00.000-08:00

When working with external files, using Cpp:
http://chishm.drunkencoders.com/libfat/
mind the makefile modifications, where to put the .a file (in the libnds folder).....and so on.....

2DArt nds

2008-10-17T06:03:00.000-07:00

2Dart nds!

can be found at:
http://dl.qj.net/2Dart-Nintendo-DS/pg/12/fid/24377/catid/221

it has been built without tiles nor sprites, it works totally drawing directly in the bitmaps
this has several advantages: for instance the letters can be drawn much more freely then using tiles or sprites....

c programming

2008-10-06T02:58:00.000-07:00

For the nds in DevkitPro we can program either in C or C++.

Here are some references for things which are a bit different in C:
http://en.wikibooks.org/wiki/C_Programming/Pointers_and_arrays

Also you have to be careful with strings, which actually don't exist in C.
char c;
char string[100]; //a string
char month1[ ]={‘j’,’a’,’n’,’u’,’a’,’r’,’y’};
char month1[]=”January”;
strcpy(mystring, "hello");//to initialize a string
strcpy(string1,string2);//copying
strcat(mystring,"hello");//got paste strings together

Character string terminated by a null character ‘\0’.

see for instance:
http://www.exforsys.com/tutorials/c-language/handling-of-character-strings-in-c.html

converting an int to a string, very basic:
u16 strokes = 12;
u8 s10 = (u8)(strokes/10);
u8 s1 = (u8)(strokes - s10*10);

char store[] = { (char)(s10 + 48) , (char)(s1+48) , '\0'};

mind the difference between " and '

Arrays:
See link on top of page.
Char string[] is already an array.
There is no evident way to get the size of an array, be careful, my trick was to put a zero at the end, and test for the zero.

Headers and Source files
You can put all the methonds or functions in the header, but accumulating headers like this has put me into trouble, so separate the method and the headers.

Don't forget to put at the top of the header file:
#ifndef __headername_H_
#define __headername_H_
and close with
#endif

otherwise you get the compiler complaining about multiple definitions of methods

Text in Tiles

2008-10-02T10:13:00.000-07:00

Liranuna used a 16x8 arrow. (see the former blog on double tiled backgrounds)
Please try in the last example an 8x8 version, and your screen will be filled with garbage!

(Use this for the gfx2gba: gfx2gba -t8 -c16 -D -P -St.bin arrowTile.bmp
and you get the .bin thing immediately)

The arrow tile number 0 fills the screen, and that is why the first 8x8 has to be transparent, then the second 8x8 has the arrow. It still works with 8x8 tiles.

This becomes clear making letters from the arrow, adding the ASCII sequence.

You need some special character sets for this, ugly and all, an 8x8 Font.

There are a few links for this:
http://www.dspassme.com/programmers_guide/tutorial/using_tiles.html

The most deep article is this:
http://www.coranac.com/tonc/text/tte.htm#sec-bmp

In the former example you can (if they are available) ask for the letters simply like this:

calculate position in array, and take place 1 of the image (the arrow or the characters)
bg0Map[POS2IDX(7, 5)] = 1 | TILE_FLIP_Y | TILE_PALETTE(2);//color 3

calculate position in array, and take place 40 of the image (the arrow)
bg0Map[POS2IDX(7, 5)] = 40 | TILE_FLIP_Y | TILE_PALETTE(2);

Since the image found in the TONC link gives the ASCII sequence, you can use the ascii value of the character:

char* msg[] = {

"http://www.dspassme.com/",
"            programmers_guide/"
};

// Draw the message on the screen.
for(i = 0; i < sizeof(msg) / sizeof(char*); ++i) {
for(j = 0; msg[i][j] != '\0'; ++j) {
sub_map[j + (i * 32)] = msg[i][j] - ' ';
}
}

(As can be seen from the dspassme link, but they dont give the whole image.)

From this starting point you can build your own characterset.
It all stays in tiles, and doesn't count on your sprites, which is quit handy because
you don't have enough sprites, if you want all our characters be present too!

So it seems like the default consoleInit, but it isn't.....we can modify!



There is one problem i encountered, the transparancy was wrong.
(Then it covers the whole screen, except for the letters.)
This was already the case with my sprites.
See the next blog for solving this, or better understanding the photoshop problem.
(It can be that new photoshops handle this without a problem, mine is old, 7.05 or so...:-)

But we can now go on: a 8x16 font, being twice as high....hmmmm what will the tiles of 8x8 do????
Well the letters are cut in half and displayed top first bottom after in the tiles, so we have
to modify the ASCII formula:


int i,j = 0;
for(i = 0; i < sizeof(msg) / sizeof(char*); ++i) {
   for(j = 0; msg[i][j] != '\0'; ++j) {
     bg0Map[j + (2*i * 32)] = (msg[i][j] - ' ') | TILE_FLIP_NONE | TILE_PALETTE(0);
   bg0Map[j + ((2*i + 1) * 32)] = ( offset + (msg[i][j] - ' ')) | TILE_FLIP_NONE | TILE_PALETTE(2);
   }
 }

Code example, 8x16 characters, click on left or right 
half of the screen to change the offset of the lower half of the tiles

tiles and backgrounds

2008-09-28T09:07:00.000-07:00

Manipulating tiles, their palettes and backgrounds asks for insight in the memory organization...

The knowledge is scattered, in simple examples, unfinished tutorials and tables. Here an effort to put it all together.

Our goal: getting a tiled background and a bitmap together, the bitmap serving as a frame around a game like chess, or minesweeper....without mixing up the memory by overlapping.

Most of the times the examples (see devkitPro) are clear while so simple. But combining some ideas immediately shows the complexity of the memory mappings the videorams and the BG registers.

The starting tile example is from Liranuna's blog: (lesson 3 and 4)

But we want to combine a bmp-background with the tiles, and we get into the trouble of mem overlapping, resulting in bad graphics.

Here a table and registers from:
http://www.dspassme.com/programmers_guide/tutorial/registers.html#bg0_cr

For the main screen:
Background Tile/Map Addresses Care must be taken in choosing the map and graphic addresses as they overlap. (Indeed!)

Address	Map Addresses	Graphic Addresses
0x06000000	BG_MAP_RAM(0)	BG_TILE_RAM(0)
0x06000800	BG_MAP_RAM(1)
0x06001000	BG_MAP_RAM(2)
0x06001800	BG_MAP_RAM(3)
0x06002000	BG_MAP_RAM(4)
0x06002800	BG_MAP_RAM(5)
0x06003000	BG_MAP_RAM(6)
0x06003800	BG_MAP_RAM(7)
0x06004000	BG_MAP_RAM(8)	BG_TILE_RAM(1)
0x06004800	BG_MAP_RAM(9)
0x06005000	BG_MAP_RAM(10)
0x06005800	BG_MAP_RAM(11)
0x06006000	BG_MAP_RAM(12)
0x06006800	BG_MAP_RAM(13)
0x06007000	BG_MAP_RAM(14)
0x06007800	BG_MAP_RAM(15)
0x06008000	BG_MAP_RAM(16)	BG_TILE_RAM(2)
0x06008800	BG_MAP_RAM(17)
0x06009000	BG_MAP_RAM(18)
0x06009800	BG_MAP_RAM(19)
0x0600A000	BG_MAP_RAM(20)
0x0600A800	BG_MAP_RAM(21)
0x0600B000	BG_MAP_RAM(22)
0x0600B800	BG_MAP_RAM(23)
0x0600C000	BG_MAP_RAM(24)	BG_TILE_RAM(3)
0x0600C800	BG_MAP_RAM(25)
0x0600D000	BG_MAP_RAM(26)
0x0600D800	BG_MAP_RAM(27)
0x0600E000	BG_MAP_RAM(28)
0x0600E800	BG_MAP_RAM(29)
0x0600F000	BG_MAP_RAM(30)
0x0600F800	BG_MAP_RAM(31)

DISPLAY_CR (0x04000000)

31	30	29	28	27	26	25	24	23	22	21	20	19	18	17	16
31	BGExtPal	29	28	27	26	25	24	23	22	21	20	FBBank		ExtMode
15	14	13	12	11	10	9	8	7	6	5	4	3	2	1	0
SprWin	Win1	Win0	Obj	BG3	BG2	BG1	BG0	BlnkScr	U6	OAMHorzBlank	ObjMapMode	3D	Mode

BG0_CR (0x04000008)

This 16 bit register controls the features related to background #0 on the main display. To activate this register, this background must be enabled in the DISPLAY_CR (0x04000000).

15	14	13	12	11	10	9	8	7	6	5	4	3	2	1	0
Size		Wrap	Map					Color	Mosaic	Tile				Pri

So, in this example:
http://www.dspassme.com/programmers_guide/tutorial/using_tiles.html#tile02

for the main screen:

POWER_CR = POWER_ALL_2D;

vramSetBankA(VRAM_A_MAIN_BG);
DISPLAY_CR = MODE_0_2D | DISPLAY_BG0_ACTIVE;
BG0_CR = BG_256_COLOR | BG_TILE_BASE(0) | BG_MAP_BASE(20);

These last three lines form a chain, combining a memory location to the videoram
tile base on 0 and map base on (a safe???) 20.

In Liranuna's example, apparently the map is smaller and put in front:
BG0_CR = BG_32x32 | BG_COLOR_16 | BG_MAP_BASE(0) | BG_TILE_BASE(1);
the map first and the tiles behind it.

Then we have to copy the tiles and palettes
memcpy((void*)BG_TILE_RAM(1), arrowTile_bin, arrowTile_bin_size);
u16* bg0Map = (u16*)BG_TILE_RAM(0);//making a pointer to the map we can calculate where to put tiles.

Adding another background? A bitmap this time, two tile-bg's would be too much :-)
vramSetMainBanks(VRAM_A_MAIN_BG_0x06000000, VRAM_B_MAIN_BG_0x06020000);

and now
either this going for 0x06010000
BG3_CR = BG_BMP16_256x256 | BG_BMP_BASE(4)  ; //this is 0x06010000
u16* backBuffer =  (u16*)(0x06000000 + 256 * 256 * 1 );

or this aiming at 0x06020000
BG3_CR = BG_BMP16_256x256 | BG_BMP_BASE(8)  ; //this is 0x06020000
u16* backBuffer =  (u16*)(0x06000000 + 256 * 256 * 2 );

is possible....

BG3_XDX = 1 << bg3_xdy =" 0;" bg3_ydx =" 0;" bg3_ydy =" 1" bg3_cx =" 0;" bg3_cy =" 0;" class="mw-headline">Map Entries
from
http://dev-scene.com/NDS/Tutorials_Day_4
of course also this example is explaining the same procedure, only it gives the tiles as a matrix.
(The last entry of this excelent tutorial.....)

The complete main.c of lesson 3 of Liranuna, but with an extra background becomes:

// Including libnds’ set of defines
#include 
// Include basic memory functions
#include 

// Including our converted graphics
#include "arrowTile_bin.h"

// Macro to translate 2D coord into 1D map array
#define POS2IDX(x, y) (x + (y*32))

// Set of defines for tile flipping
#define TILE_FLIP_NONE (0 << reg_powercnt =" POWER_ALL_2D;" display_cr =" MODE_5_2D" bg0_cr =" BG_32x32" vram_a_cr =" VRAM_ENABLE" bg0map =" (u16*)BG_TILE_RAM(0);" bg0map =" (u16*)BG_MAP_BASE(0);" bg3_cr =" BG_BMP16_256x256" backbuffer ="  (u16*)(0x06000000" bg3_cr =" BG_BMP16_256x256" backbuffer ="  (u16*)(0x06000000" bg3_xdx =" 1" bg3_xdy =" 0;" bg3_ydx =" 0;" bg3_ydy =" 1" bg3_cx =" 0;" bg3_cy =" 0;" colormask =" 0x1F;" iy =" 0;" ix =" 0;" iy =" 186;" ix =" 0;" iy =" 0;" ix =" 240;" iy =" 0;" ix =" 0;">which gives the tiles in a frame of red, located on another background layer

Liranuna gives a lesson 4 exemple with two backgrounds.
Very usefull.
I redesigned the frame exemple above with the arrows and a tiled image background.
Because the arrows are 16 colors and this tiled background 256 colors i had to shift
the colors of the tiled background, using gfx2gba, shifting 64 = 4 x 16 colors.
//gfx2gba -c256 -St.tls -t8 -St.tls -C64 -v0 -pback.pal -m back.bmp
then loading also shifted:
void *paletteBuffer =  (void *)&BG_PALETTE[64] ;
(and dmaCopy(paletteData, paletteBuffer, paletteSize); of course)
This is how the palette looks, with four 16 color palettes and behind it the colors of the tiled background:

This is the 256 color map at 0x0610000, = BG_TILE_RAM(4), and map BG_MAP_RAM(1) = 0x06000800
arrows at: BG_TILE_BASE(3) = 0x0600C000 , tiles and BG_MAP_BASE(0), map (see table above )




These examples about background organization can be found here:

Above example,using the Liranuna's arrows and adding a red frame
Second example, 2 tiled background 16/256 color (arrows and image) example

If you wonder why Liranuna used a 16x8 arrow in his example, read the following blog!

PS, so this works for the main screen, but don't think that you can copy this frame right away to the sub screen. The subscreen mem is much smaller, as you can see further on, of course an image can be displayed, as can be text, even from your own 16x8 font, but the red frame poses more problems!

Main backgrounds: 0x06000000 - 0x06400000
Sub backgrounds: 0x06020000 - 0x06021000 which is much smaller

This is a useful link with all kinds of register definitions, and here you can see the differences between VRAM_A VRAM_C and for instance VRAM_B:

http://takuto.info/libnds/arm9/video_8h.html#3bce26742d17eb8c77cf7338089f3ccf7cb3c69e848e7b5f045e932ee1e41f94

So the frame drawing (in the example red) on a bitmap can be done on the MAIN in mode 5.
double tile - backgrounds can be done, both on MAIN and SUB, in mode 0

on the SUB drawing on a bitmap (the red frame example) does not seem to be possible, in mode 5?
Yes, this is possible, but you can only use one background (apparently)

This is the example:
simple script to see drawing on both screen in mode 5

Usefull scripts

2008-09-25T06:28:00.000-07:00

Programming around for the nds is fun.
Sometimes you need something like "how to test if a point is within a triangle".
Well you can start thinking yourself, which is not bad, but other people have done it before, and still other people have refined their methods, so sometimes it is nice to use these tested scripts and go on with the game.
I found this site:
http://gmlscripts.com/

you have to rewrite the script in the language you are programming in, but that is easy :-)

ok for my problem, the triangle and the point:

{
var x1, y1, x2, y2, x3, y3, x4, y4, a, b, c;
x1 = argument[0];
y1 = argument[1];
x2 = argument[2];
y2 = argument[3];
x3 = argument[4];
y3 = argument[5];
x4 = argument[6];
y4 = argument[7];
a = (x1 - x4)*(y2 - y4) - (x2 - x4)*(y1 - y4);
b = (x2 - x4)*(y3 - y4) - (x3 - x4)*(y2 - y4);
c = (x3 - x4)*(y1 - y4) - (x1 - x4)*(y3 - y4);
return (sign(a) == sign(b) && sign(b) == sign(c));
}

this method uses three line-equations in an elegant way and tests the side of the lines, then the hypothesis that if two pairs of signs are the same, then the third has to agree, very funny!

Some info on the math.h functions is also usefull, but can be found practically everywhere.

nds Pinball 1 lines and points

2008-09-19T07:18:00.000-07:00

Just fun, this image seen at www.gooncity.com

We tried to make PINBALL on the nds. (or a minigolf, whatever takes a ball to bounce.)
A good exercise to reduce calculations and use the most simple numbers, as few doubles as possible!
The first algorithm is just a simple (naïve) reflection idea.
The ball has a position and a speed, and every time we calculate we advance the ball with its speed. We see if the ball crosses a line, if so, calculate the reflection.

We needed a point (2D) and a line struct.

The POINT struct:

typedef struct intPoint{
double x; //these members have to become int's to save space and calc's!)
double y;
}POINT;

with a few useful methods like:
dot
length
distance
checkPointBetween (just see if a point lies between two other points

nice was to discover that the sqrt (of math.h) does not function under the devkitpro with structmembers,
so we had to write our own sqrt algorithm (haha!). Found this:

double  mySqrt (double value)
{
const double tol = 0.000005; // relative error tolerance
double oldval, newval;

   oldval = value; // take value as first approximation
   newval = (oldval + value/oldval)/2;
   while (fabs((newval-oldval)/newval) > tol)
     {
       oldval = newval;
       newval = (oldval + value/oldval)/2;
     }
  return newval;
}

For the line:
typedef struct Line{
POINT p1;
POINT p2;
double cx ;//these members have to become int's to save space and calc's!)
double cy ;
double cc ;
}LINE;

with methods like:
reflection
intersection
projection
(using only geometrical methods)


All went well, (after a lot of debugging) but one can simple see that
the ball can always escape if just colliding on the point between to lines of a side.

Two escape route scenarios:



The solution might be to look after reflection if the line between the old ball position
and the new ball position crosses a line. If so, reverse speed...

But there is another way to tackle the problem, using triangles.
The field is partitioned in triangles.
The ball is always in a triangle, so only three lines to deal with.
Calculate the dcrossing and determine if the side is closed or open. If open the ball goes to the new trangle, if closed we get a reflection.
Only problem: it is too fast, (haha). Because only crossing points are calculated we have to restrain the ball to get an even movement.

Screenshot from a Devcpp prog:(with a graphic window)


Problem: if the ball is moving right on a point. We have to reflect it back.

Then we need a way to generate the triangle information.

we found Delaunay triangulation, together with some nice JAVA progs:
http://www.cs.cornell.edu/Info/People/chew/Delaunay.html
http://goanna.cs.rmit.edu.au/~gl/research/comp_geom/delaunay/delaunay.html

to be continued

Trying to describe a bug

2008-08-31T09:20:00.001-07:00

Something does not compile.
Most of the time, your own fault!

Once in a lifetime: a bug in the language.
Here is one:
but how difficult to describe it, because it is a weird combination i used, probably did it the first time ever.....

in C:

first i made a struct, just a POINT, two struct members
then declared a global instance of this struct
then i tested the sin function with a number (hardcoded) like .787878788 inside iprinf
then i tested the sin function with a structmember

the code is totally stripped:

#include --the math lib--

//this can be in a header file
typedef struct point{
double x;
double y;
}POINT;

//-------------------------

//source file:
//include header file

POINT q1;
//this global does not functions in combination with iprintf
//without iprintf no problem

int main() {

q1.x = 0.78787878787878;
q1.y = 0.686868686868686;

iprintf("point %d\n", (int) ( 100*sin( 0.12345 ) ) ) ;
iprintf("point %d\n", (int) ( 100*sin( q1.x ) ) ) ;
//multiply by 100 because it should run on the nds, without floats

return 0;
}

compile error!

declare this struct instance locally: no compile error

without the iprintf (global struct instance again): no compile error.
programming sneaky: global POINT q, local pointer to q, then iprint and sin of this pointer:
compile error!

the same in Cpp: no compile error.

ok tried everything, i agree: this is a strange case! But it seems to be real.

nds and structs in C

2008-08-28T04:34:00.000-07:00

Just an example:

I want to do several things with points and lines, so i need a few structs:

then this goes into a headerfile:

typedef struct point{
 double x;
 double y;
}POINT;

typedef struct normalPoint {
  double px;
  double py;
  double pc;
}NORMALPOINT;

typedef struct pointDouble {
  struct point p;
  double angle;
}POINTDOUBLE;

typedef struct line{
 struct point p1;
 struct point p2;
 struct normalPoint np;
 double angle;
}LINE;

and in the source file i can declare like this:

struct point q1;//or POINT q1 etc
struct point q2;
struct point newPoint ; //used to get the reflectionPoint
struct point reflectPoint ; //used to get the reflectionPoint
struct point speedPoint ; //for the speed

struct line lineArray[10];
struct line firstLine;

and use it:

q1.x = 20.0;
q1.y = 30.0;
q2.x = 120.0;
q2.y = 130.0;

firstLine.p1 = q1;
firstLine.p1 = q2;

lineArray[0] = firstLine;

then i tested the sin by printing it on the ds screen and ran into trouble!
Bad trouble indeed! See next blog.

But this has nothing to do with the structs...

Good to know: ARM9 can work with doubles, ARM7 not!
To see if the double works, on the nds screen you have to print it as an int, multiplied, (for instance) by 1000....
iprintf(" test double %d", (int)(100*sin(.787878)));

Difference between C and C++ structs (and classes): see
http://carcino.gen.nz/tech/cpp/struct_vs_class.php

Different!

2008-08-28T04:23:00.001-07:00

This is a brief recapitulation of the differences between "normal" programming and programming in the Dev-Kit pro, in C.

The pictures ->using gfx2gba see this blog, several entrees
The sounds -> using Sox, see http://www.double.co.nz/nintendo_ds/nds_develop4.html

from JAVA (or other higher languages) to C, common C errors: http://www.drpaulcarter.com/cs/common-c-errors.php
-> mind the doubles and the floats! special for the nds!!!
-> mind the sin, cos, abs, atan! Special for the nds!!! (ok, turns out to be a strange bug in iprintf)
-> mind the pointers , http://boredzo.org/pointers/
-> mind the structs http://www.learn-programming.za.net/programming_c_learn09.html
-> mind the headers!
-> mind the order in which you introduce methods!

then also because of C, well...
the makefile
the bat file

and nds pecularities:
the IRQ
the videomodes
sprites or double buffering

and debugging, having to build, having to download, putting the .nds file in an emulator,
and even that can be different from really trying your game on the nds itself!

Programming environment, not trivial either!
For me there are two possibilities:
Shared Node by Peter van Vliet available for students from HSL Leiden.
Microsoft's Visual C++ Express, specially prepaired for the DevKit Pro.

splendid homebrew

2008-07-08T06:18:00.000-07:00

Under this title a few of my favorite homebrew products
WOW! Great stuff guys!

The physics engine: http://pp.mnfu.com/
incredibly funny, must be even more fun with the DS motion
you can search on other people's idea's on the web!

Then there is DSChess: can be downloaded here: http://www.vonloesch.de/dschess
For this size of machine a very good chess player, you have to have experience to beat it and even then every move can be your last.
Seen a few bugs though: i could castle my king while being checked!
One time i could not castle although it was allowed.
Maybe it could give up sometimes when totally lost?
Minor inconveniences!!!

I like the NitroTracker too: http://nitrotracker.tobw.net/
Music can be recorded, or voice and put in a timeline, so making a rithm. Even MIDI!

there is more on MIDI and the ds here, http://dsmidiwifi.tobw.net/index.php?cat_id=1
but i could not get it wirking, probably my Firelink card (?)

Then sometimes I use DSOrganize: http://www.dragonminded.com/?loc=ndsdev/DSOrganize
very funny, i bought the webbrowser, but this prog has all installed! And even make sketches and save some notes.

There is a lot more, some are working on my firelink, some don't.
But indeed, with all these free progs around, and even source code! Who needs more?
Thx homebrewers!

I saw this page with nice games:http://www.mrbass.org/nintendoDS/homebrew/

registers

2008-07-07T12:02:00.001-07:00

The registers are quite interesting, seen as a table.
from all kind of sources these tables are collected:

first here a list with registers and addresses can be found:
http://www.dspassme.com/programmers_guide/tutorial/registers.html
in this page tables are given too, with explanations

DISPLAY_CR (0x04000000)

31	30	29	28	27	26	25	24	23	22	21	20	19	18	17	16
31	BGExtPal	29	28	27	26	25	24	23	22	21	20	FBBank		ExtMode
15	14	13	12	11	10	9	8	7	6	5	4	3	2	1	0
SprWin	Win1	Win0	Obj	BG3	BG2	BG1	BG0	BlnkScr	U6	OAMHorzBlank	ObjMapMode	3D	Mode

//-------other sources------------------------------------------------------------------

BGx_CR:(background)

b15

b14

b13

b12

b11

b10

Size

Wrap

Map Block Num

Color Mode

Mosaic

Tile Block Num

Priority

http://edorul.ifrance.com/french/chapitre12.htm

//-------------------------------------------------------------------------------------------

Attribute : 0 (sprite attribute)

15	14	13	12	11	10	9	8	7	6	5	4	3	2	1	0
Shape		C	M	Alpha		SD	R	Y-coordinate

http://home.no/neogeo/HOVEDSIDE_INDEX/GBA_HOVEDSIDE_INDEX_ENGELSK/GBA%20HTML%20LEKSJONER%201-5/GBA_Lek4_english.htm

Attribute : 1

15	14	13	12	11	10	9	8	7	6	5	4	3	2	1	0
Form		VF	HF	Rot data index			X-coordinate

Attribute : 2

15	14	13	12	11	10	9	8	7	6	5	4	3	2	1	0
Palette nummer				Priority		Sprite number

//-------------------------------------------------------------------------------------------

REG_MOSAIC @ `0400:004Ch`
F E D C	B A 9 8	7 6 5 4	3 2 1 0
Ov	Oh	Bv	Bh

http://www.coranac.com/tonc/text/gfx.htm#sec-mos

//-------------------------------------------------------------------------------------------

IRQ ( )

24	23	22	21	19 20	17 18	18	16
IRQ_WIFI	IRQ_SPI	IRQ_LID	IRQ_GEOMETRY_FIFO	IRQ_CARD IRQ_CARD_LINE	IRQ_FIFO_EMPTY IRQ_FIFO_NOT_EMPTY		IRQ_IPC_SYNC
14	13	12	11 10 9 8	7	6 5 4 3	2	1
	IRQ_CART	IRQ_KEYS	IRQ_DMA0 IRQ_DMA1 IRQ_DMA2 IRQ_DMA3	IRQ_NETWORK	IRQ_TIMER0 IRQ_TIMER1 IRQ_TIMER2 IRQ_TIMER3	IRQ_VBLANK IRQ_HBLANK IRQ_VCOUNT

http://takuto.info/libnds/arm7j/interrupts_8h.html#e384b89225fb3517bba35d3dc790c71638af5b8afc08fd0c0423b74f1e944ff9

pictures of games

2008-07-06T22:31:00.000-07:00

Doing the technical stuff is one thing.
Making pictures for the nds is another speciality.

here a link with some examples (and not to much pub-noise):
http://www.dstouch.com/

http://www.gamersunderground.net/ds/screens/

to make these pictures you have to be a real designer, photoshopper, blender specialist!

sprites attributes

2008-07-04T01:31:00.001-07:00

Using the sprites engine all kinds of attributes can be set:

most info can be found on the pages about the gba, the memory registers have stayed the same apparently.

for instance the size:
ATTR1_SIZE_8 --actually this is the tile size
ATTR1_SIZE_16 available 64 sprites, when rotscale (ATTR0_ROTSCALE_DOUBLE) is double, 32 when not double
ATTR1_SIZE_32 16 sprites in rotscale is double, or 32,
ATTR1_SIZE_64 4 sprites only, rotscale is double, or 8

then there is the color palette: (watch the number behind ATTR)
ATTR0_COLOR_256 one palette for all colors
ATTR0_COLOR_16 different palettes for sprites possible

it is possible to link a sprites (or more) to a rotation, before that you have to enable this:
ATTR0_ROTSCALE
ATTR0_ROTSCALE_DOUBLE the difference between these two is that ATTR0_ROTSCALE rotates the sprite inside the square it uses when not rotated. Double allows for full rotation. But the consequence is that the sprite is also shifted a sprite half to the right under corner. So coordinates must be revised when changing from ATTR0_ROTSCALE to ATTR0_ROTSCALE_DOUBLE. The ATTR0_ROTSCALE_DOUBLE seems to be the most logical choice, but you have less sprites. For instance for letters which will not be rotated, don't use ATTR0_ROTSCALE_DOUBLE.

ATTR0_MOSAIC you can "pixelize" the tiles: an example (background and sprite)
http://www.coranac.com/tonc/text/gfx.htm#sec-mos
but the setting of the ATTR is not enough you have to give the size of the "big pixels"
you have one possibility, one register for MOSAIC effects:
REG_MOSAIC =

0x0400 004Ch

now you have to be aware of the peculiarities of registers:
http://dev-scene.com/NDS/Tutorials_Day_3#What_is_a_register

the PA lib comes with a method to set the X and Y values of the mosaic attribute
http://www.aaronrogers.com/nintendods/PA_Lib/Day10_Mosaic/
as usual in KODERS you can find the source of these functions
http://www.koders.com/c/fidC065D03E3601EFB57A73B083E0C080699B1F4741.aspx

from all this comes the following to get the MOSAIC on sprites

int h_size = 4;
int v_size = 3;
#define REG_MOSAIC_MAIN (*(volatile unsigned int *) 0x0400004C)
#define REG_MOSAIC_SUB (*(volatile unsigned int *) 0x0400104C)
REG_MOSAIC_MAIN &= (255 <<>

useful links are coming from the gba:

http://www.webbesen.dk/gba/sprites.asp

http://home.no/neogeo/HOVEDSIDE_INDEX/GBA_HOVEDSIDE_INDEX_ENGELSK/GBA%20HTML%20LEKSJONER%201-5/GBA_Lek4_english.htm

sprites how to prepare

2008-06-20T09:56:00.000-07:00

to prepare the 16 color palette and the tiles of a 16 x 64 (4 times 16x16) sprite use:
gfx2gba -c16 -D -T16 -t8 -a0 -pgameSprites.pal -St.tls gameSprites.bmp

gameSprites.bmp is saved with indexed colors and the number of colors is not over 16.

the image is 16 pixels in height and n x 16 pixels in width (for several sprites of 16 x 16 )

c16 means 16 colors if you want to use the 256 palette use c256
T16 means the sprite is 16 wide
t8 means the tiles of the sprites are 8 x 8
a0 means the transparent color is the 0the color

gives 2 files gameSprites.pal and gameSprites.tls
64 sprites of 16x16

for 32 x 32 spites:
wide images have to be well prepared: this simple sequence might be useful:
0 32 64 96 128 160 192 234 256 288 320 352 384 416 448 480 --------- 512 544 576 608 640

to prepare the palette and the tiles of a 16 x 64 (4 times 16x16) sprite use:
gfx2gba -c16 -D -T32 -t8 -a0 -pgameSprites.pal -St.tls gameSprites.bmp

T32 means the sprite is 32 wide

only 16 sprites of 32 x 32 in a normal sprite setting

for 64 x 64 sprites:
wide images have to be well prepared: this simple sequence might be useful:
0 64 128 192 256 -------------320 384 448 512 576 640
atlhough you can only put 4 of these big sprites in normal sprite mem!

to prepare the palette and the tiles of a 16 x 64 (4 times 16x16) sprite use:
gfx2gba -c16 -D -T64 -t8 -a0 -pgameSprites.pal -St.tls gameSprites.bmp

T64 means the sprite is 64 wide

only only only 4 sprites of 64 x 64 (unless extra mem is used)

nds and the atan

2008-06-09T08:25:00.000-07:00

Hm, I wanted to shoot an arrow, in the direction in which a sprite is dragged, using the touchscreen of course.
So the arrow sprite has to be rotated, but we have the x distance and the y distance of the dragging.
To get the angle, normally the atan or atan2 is used (this even workd in Second Life!!!)
But apparently not in the nds world...
atan gives an error..... not found in devkitpro???

(ok i know there is no sin or cosin so this could be expected - look up tables is the magic word.)
(where can i find the look-up tables of the atan?)

i searched and found an interesting formula (approximation of the arctan)

((mX + 0.43157974*mX*mX*mX)/(1 + 0.76443945*mX*mX + 0.05831938*mX*mX*mX*mX)/2/PI * 512

ok: the 512 being the "degrees" on the nds ^^

of course this is only valid on domain [-1,1] for mX so we have to shift around the circle a bit....

and finally i got the function working:

int myAtan(float y, float x)
{
if ( fabs(x) > fabs(y) )
{
if (x > 0){
float mX = -y/x;
return (int)((mX + 0.43157974*mX*mX*mX)/(1 + 0.76443945*mX*mX + 0.05831938*mX*mX*mX*mX)/2/PI * 512 + 256+ 128 + 1024) % 512;
//return 0;
}
else
{
float mX = -y/x;
return (int)((mX + 0.43157974*mX*mX*mX)/(1 + 0.76443945*mX*mX + 0.05831938*mX*mX*mX*mX)/2/PI * 512 + 0 + 128 + 1024) % 512;
//return 0;
}
}
else
{
if (y > 0){
float mX = x/y;
return (int)((mX + 0.43157974*mX*mX*mX)/(1 + 0.76443945*mX*mX + 0.05831938*mX*mX*mX*mX)/2/PI * 512 + 256 + 1024) % 512;
return 0;
}else{
float mX = x/y;
return (int)((mX + 0.43157974*mX*mX*mX)/(1 + 0.76443945*mX*mX + 0.05831938*mX*mX*mX*mX)/2/PI * 512 + 0 + 1024) % 512;
}
}
}

yes fabs() is present, haha!

in dragging a sprite around another sprite you get the differences in position x, and y:

distX = touchXY.px - buttonXa; // x position of the dragged sprite - original position
distY = touchXY.py - buttonYa; // the same for the y

stupid enough ( i have to change this) i switched the x an y around in the function call:

int myAngle = (int)myAtan(positionX, positionY);

but it gives the angle and an arrow (giving power by dragged distance) displays the right direction and can be fired by letting go of the dragged sprite....

the dragging of the sprite also posed some problems, which will be documented in another blog...

of course after doing this, i tried the atan again, and it worked......
all written above is discarded, haha!!!

nds and the timer

2008-06-06T07:17:00.000-07:00

Timers are important in games, steering animations, limiting actions.
in Director there is the actionlist, and the stepFrame handler in parent scripts
in Java you have the Thread()
for the nds, all is a bit different, but in the end probably easier (?)

//----------------------first remarks, hardware timers and interrupts -----
You start with interrupts the irq, this is the basis for getting timed actions.
Then you discover the 4 timers the nds has....a timer is a set of registers hooked on the interrupt, it goes like this:

TIMER0_CR = 0;
TIMER0_DATA = TIMER_FREQ(5);
TIMER0_CR = TIMER_ENABLE | TIMER_DIV_1 | TIMER_IRQ_REQ;
irqSet(IRQ_TIMER0, yourHandler); //hooking the handler to the interrupt
irqEnable(IRQ_TIMER0);//enabling this special interrupt.

you can set a handler (called "VoidFunctionPointer") to each interrupt mask, not only timers.

The nds has for hardwaretimers for each screen, this link tells all about these timers:
http://www.double.co.nz/nintendo_ds/nds_develop8.html

Problem with these timers is that the wi connection uses them too, so in your game these timers must either be used with care or not at all.

http://www.double.co.nz/nintendo_ds/nds_develop2.html

So better is to use the VBLANK interrupt of the nds.
You have to use a method based on swiWaitForVBlank()

start the interrupts:

void irqInit();
two ways to associate a handler with the interrupt.

void irqInitHandler(VoidFunctionPointer handler);

void irqSet
(IRQ_MASK irq, VoidFunctionPointer handler;//this is the hooking one

the last one triggers on

IRQ_VBLANK
so:

irqSet(IRQ_VBLANK, yourHandler);

activate
void irqEnable ( IRQ_MASK irq );
stop
void irqDisable ( IRQ_MASK irq );

(but you need the timer or interrupt all the time, so why stop it?)

i didn't find too much docs on this subject: (in french)
http://www.playeradvance.org/forum/showthread.php?t=6103
some in english:
http://osdl.sourceforge.net/main/documentation/misc/nintendo-DS/homebrew-guide/HomebrewForDS.html 
and why not? here to practice your spanish: (interesting example included)
http://www.elotrolado.net/hilo_ds-ds-code-1-quot-interrumpiendo-quot_579855

so i tried this, having several sprites to rotate:

irqSet(IRQ_VBLANK, NULL);
irqClear ( IRQ_VBLANK );
irqSet( IRQ_VBLANK, yourHandler );//

yourHandler being the things to do every time

irqEnable( IRQ_VBLANK );

but then the movement was incredibly slow....about 3 times a sec?

fiddling around i got this solution: also start the HBLANK interrupt:

irqEnable( IRQ_VBLANK );//only enable the VBLANK -  HBLANK only did not work....

irqSet(IRQ_HBLANK, NULL);
irqClear ( IRQ_HBLANK );
irqSet( IRQ_HBLANK,

yourHandler

);//hook function to HBLANK
irqEnable( IRQ_HBLANK );

with HBLANK set after enabling VBLANK the movement was incredibly fast, so slowing down things in the movementhandler:

u8

timerCounter1 ; //a global entry, or static (whatever language you are used to)

u8

timerCounter2 ; //another one

void yourHandler

()
{
 if ( ++timerCounter1% 32 == 0) //

timerCounter being a u8 divided by 32 gives the right speed for me, experiment!

    {
//do your stuff here
}

if ( ++timerCounter2 % 16 == 0) //this one is 2 times as fast, you can go a lot faster if needed!

    {
//do your fast stuff here
}

}

now you have control over speed of movement.

//-----------------second part of the remarks about timers

Quite important is to realize how to use the timers.
In DIRECTOR you have the actorList, which gives the possibility of parentscripts using the function on stepFrame, these scripts have to be put in the actorList to start and removed from this actorList to stop.
in FLASH you can use the timer, this class just gives a signal when finishing the indicated period of time (you can start it again if necessary.)
In Java threads are used.

On the nds, you have this interrupt timer, and you have to hook in, in this process, and leave it.
This requires that you organize your processes.

for instance:

rotation a of sprite steered by the timer/interrupt,
hook this process to this timer when starting the game

when the player reaches a certain state, get the rotation of the timer
but don't stop the timer. (Other rotations of movement might still need it...)

different frequencies has to be obtained using the % function:
if (++counter % 5 == 1) only fires one time in 5

so thinking becomes different, although you can quickly organize a "timer array" with all the speed you need and let some function check this array.

again: the nds lets you get to the basis of concepts and gives you the freedom of assembling the things your way, from all kinds of basic elements.

sprite details and rotations

2008-06-02T00:29:00.000-07:00

All in the nds is a bit different from what I expected.
Ok, if you have solved the puzzle it becomes clear that indeed it is what the tuto's and the examples "say". But you have to understand the words in their context. And you can understand it differently depending on what you thought to understand from other programming environments.

So i hope to "say" things a bit more clear, for people starting programming the nds, coming from DIRECTOR or FLASH kind of environments.

So again: sprites are separate from their palettes and now it turns out separate from their rotations!

Example: Say, you have a few sprites, 16 x 16 pixels, 256 colors (so one palette for all)

Then you load these sprites as a sequence: (see also above)

this is an image consisting of 11 sprites: (of course a bit larger then for the nds, :-)

the magenta colour counts as transparent, and the "replay" text consists of three sprites...

You have to initialize it like that:

setSprites(SCREEN_SUB, &gameSprites_tls, gameSprites_tls_size, &gameSprites_pal, gameSprites_pal_size);
(depending on your method, you have to provide screen number, the tiles, tileSize, palette, paletteSize....)

(but the main thing is: all "sprites" are put in at once, this is: put in memory, not yet on the screen.)

now you have to put a piece of this long thing on the screen as "a sprite":

setSprite(SCREEN_SUB, 0 , 0, 0, 200, 10 );
setSprite(SCREEN_SUB, 1 , 1, 0, 200, 30 );
setSprite(SCREEN_SUB, 2 , 1, 0, 200, 50 );
setSprite(SCREEN_SUB, 3 , 2, 3, 200, 70 );
setSprite(SCREEN_SUB, 4 , 2, 2, 200, 90 );

where
the first indicates the screen,
the second: the number (actually in the OAM) of your sprite to be
the third is the index of which rotation is linked to this sprite, some are linked to the same rotation!
the fourth: which piece of the long strip is going to make it to be your sprite
fifth, sixth: x and y

now if you want to play with your sprite: moving it:
you have to refer to the sprite number (inOAM) so this is the second number:

moveSprite(SCREEN_MAIN, 3 , posX, posY ); //or methods like this
so here i move sprite 3, to position posX, posY

but watch carefully now:
rotating the same sprite:
rotateSprite(SCREEN_MAIN, 2 , rotX );
but since sprite 4 is linked to this rotation, it will move to!!!!
actually you change the rotation number 2, and with this the sprites which are linked to it.

so rotateSprite, as a name is not very well chosen: better changeRotation, or something....

Makefile monsters

2008-06-01T00:37:00.000-07:00

Having the incredible urge to make a game for the nds, you dont think of the makefile.
You think of your game!

After having sniffed at all sorts of examples you try something of your own, and "Fi Donc!", (french for all kinds of trouble!), the makefile doesn't quite let you taste the sweetness of succes....:-(

What is this monster? This is a file with rules how to link all the things you have put together in your project, and, finally presenting the code, the images, the sounds to the devkitpro, to let it be crunched and digested. All of which has to be done to be able to produce the miracle .nds file, which can be put in the emulators or the nds itself.

Wow!

You can do this by hand too!
http://www.dev-scene.com/NDS/Tutorials_Day_2#Build_A_Demo_The_Hard_Way
the name of this link speaks for itself.

There is some luck around the corner: you are not the first to confront this....
Killroy was here! is written on this wall, and whoever it was, they were good!!!
They have created templates for the makefile.

Want to see a not so template makefile?
Look at this sound example:
http://www.double.co.nz/nintendo_ds/nds_develop4.html
and download and open this makefile, it shows its working very clearly, totally dedicated to this very instructive sound project!

Then go to the devkitpro examples, there you see more generic makefiles, although the methods to work with sprites differ, so, the even the generic makefiles will differ amongst themselves.

in between and writing about the "old" style and the "new" is this link:
http://www.fampennings.nl/maarten/nds/index.htm
here the workings are explained fixing the watch example in de Devkitpro examples in the meantime, and expanding on the two processors ARM7 and ARM9.

Now that you now the why and the how, you can go and talk to these monster makefiles...
what helps to is searching a tutorial on makefile, although this tends to be too general, you can extract some rules and get used to the very short commands....
http://www.opussoftware.com/tutorial/TutMakefile.htm

I use C++ Visual Express for the moment to compile my test projects:
page of devkitpro explaining how to set up:
http://www.devkitpro.org/index.php?s=cpp
(yes, it must be a makefile project)
In the project folder has to be placed the makefile.
In my experience, with projects with a few sprites sounds, and images for backgrounds, the compiler always forgot "something". Cleaning the project didn't help. So i didn't clean anymore, just where the compiler provided errors on images or sounds, i removed these files (all of them .o, .bmp etc of the same name) and generated the project again, this saves compiling time and finally it worked.
Why didn't it work either at once, or not? Computerprogramming is an exact science, no? No comprende! (Spanish for all sorts of trouble!) Maybe my RAM is not sufficient, maybe devkitpro is strange????

later on (in nds traditions never) there will be more on makefiles!!!

this is the offcial description:
http://www.gnu.org/software/make/manual/make.html

Sound

2008-05-31T23:57:00.000-07:00

What is a game without sound? What is anything without sound?
What is even (our experience of) space without sound???

So sound is important.
But first of all, you want to have pictures, movement, and or levels in your game.
Ok, if you are on your way with this sound has to be added.

Difficult?

Hmmmm, as everything on the ds, somehow difficult, after a while less so.

16 channels!

The first "layer" is not too difficult, but it can become complex, just because also sound is complex, not just sprites!

links:
the simple devkitpro example: sound is very instructive and shows how easy it is, it gives two examples in one:
see all code there:

//first way setting up:
// set the generic sound parameters

setGenericSound(
11025, /* sample rate */
127, /* volume */
64, /* panning */
1 ); /* sound format*/

//second way:
TransferSoundData blaster = {

blaster_raw, /* Sample address */
blaster_raw_size, /* Sample length */
11025, /* Sample rate */
127, /* Volume */
64, /* panning */
1 /* format */
};

both are simple to implement and understand:

if ( keys & KEY_A) playGenericSound(saberoff_raw, saberoff_raw_size);

//using first setting up
if ( keys & KEY_R) playSound(&blaster); //second

Preparation:
the sounds have to be prepared in sox, from a .wav fromat to a .raw format, can be included like that and it works!
but then you need a loop sound!!!

http://www.double.co.nz/nintendo_ds/nds_develop4.html

Here you can read more about the sound settings, and see that it might become a bit more complex!
difference between one shot and loop!

http://nocash.emubase.de/gbatek.htm#dssound

Here apparently are the registers the sound uses, and then you can really master sound!

Another detail: (why keep it simple?)
there are two processors, the more complex examples shift the sound control to the ARM7, instead of the game which is located on the ARM9. This asks for an extra line in the makefile monster of course....more experience needed!

a link we met before does a lot of explaining, in section 4.1 - 4.4 :
http://www.fampennings.nl/maarten/nds/index.htm

But what a minute:
You can record sound, or use the microphone actively in a game!
the last links next entry is the DS Sound Capture. FUN!
(Let the player get her next levels only after shouting in the mic! )
This stays an idea for the moment!

in great nds tuto traditions, we leave the remarks on sounds in this unfinishe state!!!!

nds sprites intro

2008-05-29T07:51:00.000-07:00

Strange: there are a few nds sprites tuto's around, and very good ones, but somehow these are also very much starters and then most of them are suddenly incomplete.
Suddenly? Yes, they leave out the palet problem, or the problem of more then one sprite, or just stop the text!

The Introduction to Nintendo DS Programming of Jaeden Amero
is an exception, but unfortunately his examples have variables in it that won't compile in my PC (such as the temporary oam: toam ). (Due to my first try in PALIB and installing not the latest version of devkitpro?)(I have to try to run it in Visuall C++ express...)

What is the problem with sprites, these beasts being so well managed nowadays in DIRECTOR or FLASH?

well, for a start they have to be put in the nds, in the memory
the ds has to be prepared for this in the videomode
you have to prepare the sprites images in certain formats
these images consisting of (tiles bits broken up) and palettes have to be included using the makefile monster.
Who did ask this makefile monster to join the party?

If coming from DIRECTOR, FLASH you have to rethink your sprites ideas:
for instance: if working with a 256b color palette, there is only one palette so what you actually construct is one big image and you pick out the sprites by an extra index.
So you load the whole alphabet as a sprite, and you get the E by indexing on 4 (starting with 0 of course). Then your sequence of letters has all the same palette.
Nice example:
http://liranuna.drunkencoders.com/misc-projects/tileoptimize/

Examples can be found in devkitpro, but if you look carefully the examples are depending on the makefile and sometimes using grit, or other similar but different progs (gfx2gba), sometimes reading in bmp files, all kinds of possibilities, but my experience was that: ok this is functioning, but why?

examples of devkitpro online for the latest updates:
http://devkitpro.cvs.sourceforge.net/devkitpro/examples/nds/

so we have to scramble sites and blogs to get an idea how this works.
for me the main problem was getting a view on the problems! (The differences with my previous encounters with sprites...)

List of problems

0. which programming environment to use: (after installing devkitpro)
1. the dependency on the makefile (functioning in the folder devkitpro...),
2. what is the grit (or others, gfx2gba, .pcx) relation, how to make the images readable for C and the nds
3. how to prepare the nds video mode (seems easy!)
4. how to get these data and palettes into the right spot in the memory of the nds

Solutions and answers:

0. which programming environment to use:
Programmers notepad is ok, but running slow and not always compiling the examples in the same way.
Visual Express C++, quite a monster but in my experience far more suited, you have to rebuild the examples using: http://www.devkitpro.org/index.php?s=cpp

1. the dependency on the makefile (functioning in the folder devkitpro...),
.... is still a bit unknown territory for me, related to devkitpro, study in progress

2. what is the grit (or others, gfx2gba) relation, how to make the images readable for C and the nds
....idem, the main problem being not grit but the dependency on the makefile, study in progress
examples in devkitpro:
Complex_2D, has all the OAM and sprite functions, reads in a pcx image
Sprite_Bitmap, makes its own (simple) sprites
some homebrew have .git and .grit files (by wingit.exe)
some homebrew use gfx2gba.exe and get .tls and .pal and .map files
some read the image immediately

3. how to prepare the nds video mode (seems easy!)
All kinds of choices for sprites you need a text mode (texture mode), for the screen with the sprites, MAIN or SUB, see (for instance)
http://www.tobw.net/dswiki/index.php?title=Graphic_modes
or the very extended (but suddenly stopping)
http://www.dev-scene.com/NDS/Tutorials_Day_2
or again on dev-scene:
http://dev-scene.com/NDS/DOCgraphicmodes
For this problem you can compare the examples and learn a lot about this Videomode.

4. how to get these data and palettes into the right spot in the memory of the nds
to get the picture of the OAM memory, look at this here (a french site but the picture is universal!) http://www.dev-fr.org/wiki/nds/doku.php?id=dssprite
you can see what 's happening with your sprite, data-part gfx, it gets split up in tiles!
but what happens with the pallette, they dont tell!

This we can read in Nintendo DS Programming of Jaeden Amero

a few very useful details about tiles and palettes, coming from this text:

There are two ways to tile sprites, 1D and 2D. In a 2D layout, the sprite memory is treated like a giant image from which sprite tiles are obtained by making a box the size of the sprite and placing it over the giant image. In a 1D layout, the sprites are layed out in a linear fashion,

Sprites can be made of tiles that use 16 colors or of ones that use 256 colors.
liranuna gives a picture, this is my own:

Tiles that use 256 colors are twice as large as tiles that use only 16 colors. Another advantage of using 16-color sprites is the ability to use 16 different palettes. When using 256-color sprites, each sprite (and tile) must use the same palette. When using 16-color sprites, we can have one sprite use one 16-color palette and another sprite use a different 16-color palette, even though both use the same tile data.

then you have to prepare the OAM, well this is everywhere the same and easy.
see again: Nintendo DS Programming of Jaeden Amero

then you have to get your spites in, this is low level, and all these functions can be put away in a sprite.c file, but there is no general sprite.c file around! Everyone makes his own, which is much more detailed then is shown in the examples.

in full tradition of all what is written on the nds:

more in blog sprites details!

text in progress

nds links

2008-05-22T12:14:00.000-07:00

just a starter blog: first all kind of things...

the start of nds programming coming from "normal" programming is a bit steep, due to the low level things with makefiles and images....

my first applications was an OpenGL thing, with models imported from BLENDER (with the export Python script)
but OpenGL on the nds is sort of special feature...
then i reshaped a program just to show pictures above and some text below, a sort of slide show, or presenter

then i bounced against the wall of the makefile
i was recoiled into sprites,
happened to stumble into tiles and maps
etc etc etc

emulators can be found...everywhere, i used DUALIS, now NO$GBA.EXE, sometimes: ideas.exe.
the best experiences lately with
NO$GBA.EXE
http://www.emulator-zone.com/doc.php/gba/nocashgba.html

nearly all tuto's on the web about nds programming are not really finished, some stopping with sprites, others (sort of like this sentence) you have to scramble and try out, the list below is about the best there is...

this is just a page where to find some readable (sometimes partial) tutorials

installing devkitpro is well covered
in devkitpro all kind of examples are given

the main problem of nds programming is the setting up, after that it becomes more and more just depending on yourideas and programming skills

so first....
grit, gfx2gba.exe, PAGfx
backgrounds, tiles, maps, sprites, palettes
VRAM, memory banks, registers, MODE
then you can start wth your game and it becomes familiar OO...

Oh yes i forgot: you have to have something to run your program on, inside the DS, i wanted to have an R4, but that was not available and opted for a fire link card, this proved to be fine, having a direct USB link to the PC, and acting like an exteral hard disc. Some programs won't work, but ok....
Memory 1 GB, seems to be a lot for these small progs.
All kinds of pages about these cards, nearly too much.....

Oh yes, most of the programming is done in C, although PALIB can be done in C++...
but making sprites and nice maps is much more difficult! The planning becomes so important!

//---------------------------------------tuto's-----------------------------------------------
homebrew (problem: info seems good but not all seems to work ...due to my fire link card???)

http://liranuna.drunkencoders.com/nds-2d-tuts/
nice soft intro to getting tiles and maps into the nds (runs inside devkitpro)
but the way it is programmed is quite different, older? then the devkitpro examples

a very worthwhile manual:
http://patater.com/files/projects/manual/manual.html

http://www.dev-scene.com/NDS/Tutorials_Day_1
of course this is very good, but this sequence stops at sprites, just where it matters most!

this seems to cover a lot, but it uses GIMP with a Python script i cannot find (for windows)
http://www.dspassme.com/programmers_guide/tutorial/

specifications
http://nocash.emubase.de/gbatek.htm

this explains why and what about sprites, and these sprites seem to be a lot of work, but fun!
http://osdl.sourceforge.net/main/documentation/misc/nintendo-DS/graphical-chain/OSDL-graphical-chain.html
coming from here:
http://osdl.sourceforge.net/main/documentation/misc/nintendo-DS/homebrew-guide/HomebrewForDS.html

then there is the PALIB
http://www.palib.info/Doc/PAlibDoc%20Fr/index.html
it has something if you can develope in C++ Visual Express
PALIB tuto: http://www.palib.info/wiki/doku.php?id=day1

//-------------------------------------openGL-----------------------------------------
OpenGL examples can be found inside devkitpro, it feels like a bit of outside world in the nds programming and is pretty much like ... OpenGL :-)

//-------------------something about makefiles------------------------

http://www.coranac.com/tonc/text/setup.htm

//------------------------------------------------------------------------------------
where to find some of the main free applications:
http://dshb.50webs.com/pages/hb.htm
http://www.ndshb.com/cgi-bin/cfiles/cfiles.cgi?0,0,0,0,12,135
i like the chess program, the dsorganize and the dslinux, for a start...

//---------------------------------------to get images in right shape for sprites or backgrounds

Usenti, grit, wingrit, gfx2gba downloads:
http://www.coranac.com/projects/#usenti

//------------------------------------------------------------------------------------
short tricks: (some things not worth a whole tuto, :-)
swapping upper an lower screen (just: lcdSwap(); )

lcdMainOnTop(); //MAIN renders to top screen SUB to bottom
lcdMainOnBottom(); //SUB renders to top screen MAIN to bottom
lcdSwap(); //swap the screens the egines were rendering to