I would like to initially set a CGPoint property to a particular point (middle of screen). Other methods may subsequently wish to change this property. My thoughts were to initialise it if empty in the getter, but I get the message invalid argument type 'struct CGPoint' to unary expression. I also tried using if property == nil or 0 but no joy.
Any thoughts?
-(CGPoint)graphOrigin
{
// initialise to centre of screen if has not been set
if(!_graphOrigin) // this expression is causing the problem
{
CGPoint origin = CGPointMake(self.bounds.origin.x + self.bounds.size.width / 2, self.bounds.origin.y + self.bounds.size.height / 2);
_graphOrigin = origin;
}
return _graphOrigin;
}
A CGPoint is a struct, so you can't set it to nil or NULL (it's not a pointer). In a sense, there's really no "uninitialized" state. Perhaps you could use {0.0, 0.0} to designate an unset CGPoint, but that's also a valid coordinate. Or you could use negative x and y values to flag an "uninitialized" point, since negative values can't be valid drawing points, but that's a bit of a hack, too.
Probably your best bet is to do one of two things:
Store the property as a pointer to a CGPoint. This value can be set to NULL when uninitialized. Of course, you have to worry about mallocing and freeing the value.
Store the CGPoint alongside a BOOL called pointInitialized or somesuch, initially set to NO, but set to YES once the point has been initialized. You can even wrap that up in a struct:
struct {
CGPoint point;
BOOL initialized;
} pointData;
An easier way would be to initialize _graphOrigin to CGRectZero and change your if statement for this:
if (!CGPointEqualToPoint(_graphOrigin, CGPointZero)) {
}
CGPoint does not have an uninitialized state. However, if you consider the point (0, 0) as uninitialized, you could use
if (_graphOrigin.x == 0 && _graphOrigin.y == 0)
{
...
This works because when an Objective-C instance is initialized, all its ivar are cleared to bits of zero, which in the CGFloat representation is 0.0.
(Note: The == is fine here even if the operands are CGFloat because we want to compare with the an exact bit pattern (ignoring the issue of -0))
Since CGPointZero (0,0) and any other value you give a point may exist in your context
you may want to initialize an NSValue with your point using:
NSValue *pointVal = [NSValue valueWithCGPoint:point];
You could do this based on some condition and then later test the NSValue for nil.
NSValue can also be added to an array which would allow you to have an array of points should you need.
To get the point later simply use:
CGPoint point = [pointVal CGPointValue];
static CGPoint kInvalidPoint = {.x = NSIntegerMax, .y = NSIntegerMax};
#implementation MyClass
- init()
{
self = [super init];
if (self) {
_oldPoint = kInvalidPoint;
}
return self;
}
- (void)foo
{
if (CGPointEqualToPoint(self.oldPoint, kInvalidPoint)) {
// Invalid point.
return;
}
}
#end
Create two CGPoint properties, that way they are both "uninitialized". Set one of them and use the second one to check whether or not they are equal.
#interface ClassName ()
#property (nonatomic) CGPoint point1;
#property (nonatomic) CGPoint point2;
#end
#implementation ClassName
self.point1 = CGPointMake(69.0f, 180.0f); //arbitrary numbers
//if not equal, then if statement proceeds
if (!CGPointEqualToPoint(self.point1, self.point2) {
//your code here
}
#end
Idk if you'd consider this way hackish though. And I know your question was already answered, but I had kinda the same dilemma till I thought of this.
Related
I wanted to create 2 global arrays which can be updated during the run of the programme.In each update i add one element to zeroth position and deleted the last number
I created the arrays as....
In the .h file..........
//////////////
#interface Shared : NSObject{
NSMutableArray *x;
NSMutableArray *y;
}
#property (nonatomic,retain) NSMutableArray *x;
#property (nonatomic,retain) NSMutableArray *y;
+(Shared*)sharedInstance;
#end
In .m file
staticShared* sharedInstance;
#implementation Shared
#synthesize x;
#synthesize y;
+(Shared*)sharedInstance
{
if (!sharedInstance) {
sharedInstance=[[Sharedalloc]init];
}
returnsharedInstance;
}
-(Shared*)init
{
self = [superinit];
if(self)
{
x=[[NSMutableArrayalloc] init];
x=[NSMutableArrayarrayWithObjects:#"0",#"0",#"0",#"0",#"0",#"0",#"0",nil];
y=[[NSMutableArrayalloc] init];
y=[NSMutableArrayarrayWithObjects:#"0",#"0",#"0",#"0",#"0",#"0",nil];
}
returnself;
}
#end
Then i used to call them and re,ove and added elements using the following code....
[[shared sharedInstance].y removeLastObject];
[[shared sharedInstance].y insertObject:new_element atIndex:0];
[[shared sharedInstance].x removeLastObject];
[[shared sharedInstance].x insertObject:new_element atIndex:0];
In the mean time i call these values and calculate an arithmetic value using an expression.
This seems to work well. But it seems to be an inefficient way to handle floating point numbers which i store in it. As these arrays creates objects. Is there any easy method that i can create a global array containing specified amount of floating point numbers and update it during the run of the programm(array size is fixed) by deleting the last object, and call them back to do calculation?
Please help me!
EDIT 1
To sir deanWombourne
.................................
I implement as you instructed! Can you please go through this and help me to correct 2 errors i get.
IN the .h file
#interface Shared : NSObject{
#private
float input[7];
float output[6];
}
+(Shared*)sharedInstance;
-(void)addNewInput:(float)input1;
-(float *)input;
-(void)addNewOutput:(float)output1;
-(float *)output;
#end
in .m file............
#implementation Shared
-(id)init{
if((self =[superinit])){
for(int n=0; n<7 ;++n)
input[n]=0.00f;
for(int n=0; n<6 ;++n)
output[n]=0.00f;
}
returnself;
}
-(void)addNewInput:(float)input1{
input[0]=input[1];
input[1]=input[2];
input[2]=input[3];
input[3]=input[4];
input[4]=input[5];
input[5]=input[6];
input[6]=input1;
}
-(float *)input {
returninput;
}
-(void)addNewOutput:(float)output1{
output[0]=output[1];
output[1]=output[2];
output[2]=output[3];
output[3]=output[4];
output[4]=output[5];
input[5]=output1;
}
-(float *)output {
returnoutput;
}
#end
When calling it
float reading= (accel_reading)/(1.165969038*1e5f);
[[SharedsharedInstance] addNewInput:reading];
Problems i get
1. In the implementation, it says incomplete implementation (it's a warning not an error)
2. How can i used a for loop to fill array values or is this way ok?
Major problem i get,
When i call it as shown above, program stops running telling
Terminating application due to uncaught exception 'NSInvalidArgumentException', reason '+[SharedsharedInstance]: unrecognized selector sent to class 0x5780'
Please help me through this...............
Your code Smells (and I mean that in the nicest possible way!)
Using two parallel arrays and keeping in sync is a bad design pattern (and a performance hit in quite a few ways!). Especially as there is already a struct that handles storing an x and y at the same time - CGPoint).
You're solving the 'only objects go in arrays' problem by converting your float' primitives toNSString` objects, which is horrendously inefficient - take a look instead at the NSValue class, it's designed to put native C primitives into an object without expensive parsing operations :)
You might also want to look into malloc (and free etc) and deal with the whole problem at the C level - this will mean no objects at all and would be blindingly fast (at the cost of more complicated code).
Hope this helps, if you have any questions just add a comment to this answer :)
EDIT
If all you want to do is store 4 x and y values, then this is probably the easiest way to do it :
#interface Shared : NSObject {
#private
CGPoint points[4];
}
+(Shared *)sharedInstance;
- (void)addNewPoint:(CGPoint)point;
- (CGPoint *)points;
#end
#implementation
- (id)init {
if ((self = [super init])) {
// Start with 0,0 for all your points
for (int n = 0; n < 4; ++n)
points[n] = CGPointZero;
}
return self;
}
- (void)addNewPoint:(CGPoint)point {
// Just move all the points along one and add the new one to the end
// (yes, this could be done in a loop but there's not that much point for 4 points!)
points[0] = points[1];
points[1] = points[2];
points[2] = points[3];
points[3] = point;
}
- (CGPoint *)points {
return points;
}
#end
This gives you a method addNewPoint that removes the first point and adds the new point to the end of your array.
You also get the method points that returns the 4 points. Use it something like :
// To add a point
CGPoint newPoint = CGPointMake(100, 100);
[[Shared sharedInstance] addNewPoint:newPoint];
// To do something with the points (in this case, NSLog them)
CGPoint *points = [[Shared sharedInstance] points];
for (int n = 0; n < 4; ++n)
NSLog(#" Point %i : %#", n, NSStringFromCGPoint(points[n]));
EDIT #2
From your comments, you need two arrays, one with input data and one with output data. Try something like this :
#interface Shared : NSObject {
float inputs[4];
float outputs[5];
}
...
This will give you two arrays to read/write to - one called inputs and the other called outputs. Access them in pretty much the same way you did the ones in my first edit :
float *inputs = [[Shared sharedInstance] inputs];
for (int n = 0; n < 4; ++n)
NSLog(#" Input %i : %f", n, inputs[n]);
float *outputs = [[Shared sharedInstance] outputs];
for (int n = 0; n < 5; ++n)
NSLog(#" Output %i : %f", n, output[n]);
Would a linked list be overkill for what you're trying to achieve? It's not quite as simple as a static array of floats, but makes the removal of the last object and insertion of the zeroth object reasonably simple and fast.
If you want an array containing a specific number of Objects, you can use NSArray, which is static, opposed to NSMutableArray.
As for the array being Global, just implement a singleton class that contains the 2 arrays and provides the associated methods.
in Globals.h:
#interface Globals : NSObject
+ (Globals *) sharedGlobals;
#end
in Globals.m:
#implementation Globals
static Globals *sharedGlobals = nil;
+ (Globals *) sharedGlobals{
#synchronized(self){
if (sharedGlobals == nil){
sharedGlobals = [[self alloc] init];
}
}
return sharedGlobals;
}
you then can access the arrays (after you implemented them) with the following line:
[[Globals sharedGlobals] getArrayX];
Here is a sketch to get you going.
Your array size is fixed and only contains floating point numbers, start with a C array:
double x[] = {0, 0, 0, 0, 0, 0, 0};
double y[] = {0, 0, 0, 0, 0, 0};
The number of elements in these arrays can be calculated rather than hard-coded:
int xCount = sizeof(x)/sizeof(double);
int yCount = sizeof(y)/sizeof(double);
Now use these arrays as a circular buffer, declare a cursor and initialise:
int xCursor = 0;
The item at the front of the queue is at the cursor:
valueAtFrontOfQueue = x[xCursor]; // get the current front item
To remove the value at front and add a new one to the rear replace the value at the cursor with the new value and increment the cursor:
x[xCursor] = newValueForBackOfQueue; // replace it with new item for back of queue
xCursor = (xCursor + 1) % xCount; // and advance cursor using mod arithmetic to it cycles around
No wrapping doubles as objects, no dynamic allocation at all.
Wrap the above up as you see fit, maybe as a class, and you're done.
I have a struct Position, like this:
typedef struct Position { int x, y; } Position;
How may I pass it in NSObject performSelector:withObject:afterDelay:? Like this:
Position pos;
pos.x = pos.y = 1;
[self performSelector:#selector(foo:)
withObject:pos // ERROR
afterDelay:5.0f];
EDIT: changed code to fix typo
Uhm.. use a CGPoint and something like
[self performSelector:#selector(foo:)
withObject:[NSValue valueWithCGPoint:CGPointMake(pos.x, pos.y)]
afterDelay:5.0f];
And read it again as
NSValue v;
CGPoint point = [v CGPointValue];
or leave the Position class away completely, CGPoint does the same
You could wrap your custom type using NSValue class. The error is that you didn't provide an object reference to the method.
Try using NSValue's +(NSValue *)valueWithBytes:(const void *)value objCType:(const char *)type; class method. On the other side you can retrieve the value using -(void)getValue:(void *)buffer;.
preformSelector:withObject: accepts only objects as parameters, hence you'll have to implement your foo: method to accept an object. There are two ways to pass the struct as an object:
create a struct-like object or
wrap it into NSValue and unwrap it in the method.
Full answer, based on user756245's (which doesn't tell you how to use it, not a great deal of help). Also, Apple suggests you use a slightly different method these days, IIRC:
typedef myStruct { int a; } myStruct;
myStruct aLocalPointer = ... assign to it
[self performSelector:#selector(myOtherMethod:) withObject:[NSValue value:&aLocalPointer withObjCType:#encode(myStruct)] afterDelay:0.25];
This is most likely asking for trouble, but you can pass CGPoint as id by bridging it in this way:
withObject:(__bridge id)((void *)(&point))
This will lead to a crash if point goes out of scope and your selector tries to read it, though.
Is there a way to make nullable struct in objective C like in C# you can use Nullable<T>?
I need a CGPoint to be null when there is no applicable value. I cannot allocate a random invalid value for this like (-5000, -5000) because all values are valid for this.
What if you define a CGPoint using CGPointMake(NAN, NAN) similar to CGRectNull? Surely with NAN's for coordinates, it's not still a valid point.
CGPoint is a struct and that has some different rules in objective-c than you might think. You should consider reading about structs in objective-c.
The way this is done most of the time is to wrap the struct in an object because that object can be set to null. NSValue will wrap a CGPoint.
NSValue * v = [NSValue valueWithPoint:CGPointMake(1,9)];
NSVAlue * vNull = [NSValue valueWithPointer:nil];
if([v objCType] == #encode(CGPoint)) printf("v is an CGPoint");
CGPoint is a enum, not an object. You can use CGPointZero, or you can wrap all of your points inside of NSValue, which are objects and can be nil.
There is also nothing stopping you creating your own struct based on CGPoint, similar to how C# 2 works.
struct NilableCGPoint { bool isNil; CGPoint point; }
Examples of use:
// No value (nil)
NilableCGPoint myNilablePoint.point = CGPointZero;
myPoint.isNil = YES;
// Value of (0,0)
NilableCGPoint myNilablePoint.point = CGPointZero;
myPoint.isNil = NO;
// Value of (100, 50)
NilableCGPoint myNilablePoint.point = CGPointMake(100, 50);
myPoint.isNil = NO;
I want to make an method which takes an CGFloat by reference.
Could I do something like this?
- (void)doStuff:(CGFloat*)floatPointer
I guess this must look different than other object pointers which have two of those stars. Also I'm not sure if I must do something like:
- (void)doStuff:(const CGFloat*)floatPointer
And of course, no idea how to assign an CGFloat value to that floatPointer. Maybe &floatPointer = 5.0f; ?
Could someone give some examples and explain these? Would be great!
objective-c is still c, so
-(void) doStuff (CGFloat *) f
{
*f = 1.2;
}
call with
CGFloat f = 1.0;
[self doStuff:&f];
If you are passing a CGFloat by reference, then accessing it is simple:
- (void)doStuff:(CGFloat*)floatPointer {
*floatPointer = 5.0f;
}
Explanation: as you are getting a reference, you need to de-reference the pointer (with the *) to get or set the value.
if you (hate pointers and ;-) prefer objective-c++ pass by reference, the following is an alternative:
-(void) doStuffPlusPlus:(CGFloat &) f
{
f = 1.3;
}
call by
CGFloat abc = 1.0;
[self doStuffPlusPlus:abc];
and, you need to rename the source filename from ???.m to ???.mm
The common Superclass of Rectangle and Circle is Shape.
If I initialize some shapes, what is a good way of converting the shape into a circle later and keeping the same properties set while it was a shape? Should I implement a initWithShape in the subclasses that looks something like this?
- (id) initWithShape:(Shape*)aShape {
self = (id) aShape;
// set circle or rectangle specific values
return self;
}
Does anyone have an example that I can look at?
Don't do what you just did. Think about what happens when you do this:
Shape *shape = ...;
Rectangle *rect = [[Rectangle alloc] initWithShape:shape];
In the second line, an instance of Rectangle gets allocated. Then, the return value for initWithShape is just shape again, so the new Rectangle that we just allocated has been leaked!
The cast to id is also unnecessary—any Objective-C object can be implicitly cast to id.
I'm not entirely clear on what you're trying to do. Perhaps if you clarified your question, I could tell you what you should be doing.
You cannot change an object after it has been created, except by freeing it and creating a new one (which you can do in an init method, and is in fact quite often done for singletons or class clusters), but that is not really what you're after.
Give an existing Shape object, with some properties, your only real option is to create a new object based on the shape properties. Something like:
In Shape.m:
- (id) initWithShape:(Shape*)aShape {
self = [super init];
if (self != nil) {
// copy general properties
_x = aShape.x;
_y = aShape.y;
_color = [aShape.color copy];
}
return self;
}
In Circle.m:
- (id) initWithShape:(Shape*)aShape {
self = [super initWithShale:aShape];
if (self != nil) {
// base properties on the class of the shape
if ( [aShape isKindOfClass:[Oval class]] ) {
// average the short and long diameter to a radius
_radius = ([(Oval*)aShape shortDiameter] + [(Oval*)aShape longDiameter])/4;
} else {
// or use generic Shape methods
_radius = aShape.size / 2;
}
}
return self;
}
If you have a reference to a Shape, and it might be a Rectangle or Pentagram or whatever, and you want to 'convert' to a circle (I guess you mean a circle with the same bounding box?), you have to create a new object. You can't change the class of an object after it's been created (except through very nasty low-level hacks.)
So yes, you would create an -initWithShape: method in class Circle. But the method would look like a normal init method, setting up the instance variables of the new Circle object ('self'). It would access properties of the given Shape, like its position and size, and set up the new object accordingly.
Why not implement a method in your shapes to take properties from other shapes rather than trying to replace the instance of the object altogether. It's probably safer.
// for rectangle
- (void) takePropertiesFrom:(Shape *) aShape
{
if ([aShape isKindOfClass:[Circle class]])
{
float radius = [aShape radius];
[self setWidth:radius * 2];
[self setHeight:radius * 2];
}
else
{
[self setWidth:[aShape width]];
[self setHeight:[aShape height]];
}
}
// for circle
- (void) takePropertiesFrom:(Shape *) aShape
{
if ([aShape isKindOfClass:[Rectangle class]])
[self setRadius:[aShape width] / 2];
else
[self setRadius:[aShape radius]];
}
Obviously you would want to set up a public interface for Shape that exposes the basic properties of a shape, such as height and width, and then you won't need to hard-code property stealing based on class type.