I know how to use structs in the traditional C way. My question is how does one develop a class (correct terminology?) using a struct in the vein of NSPoint, NSRect, etc? Say, for example, I want a 4-D point:

struct PointInSpaceTime { double x; double y; double z; double t; }; typedef struct PointInSpaceTime PointInSpaceTime;

And I want a method that operates on this struct.

-(BOOL) isInPerfectHarmony; { if ((x==y) && (y==z) && (z==t) return TRUE; else return FALSE; }

I have read in other places to wrap the struct in NSData (or NSValue depending on the responder). If this is a suggestion, how do I implement this example (code example, please).

Thanks for any responses! Any information would be beneficial!

Philip D Riggs

There are a million different ways to to do this. Well - probably not a million… certainly several… ;)

Here’s one…

@interface PointInSpaceTime { @private double x; double y; double z; double t; }

//these methods are a convenience for constructing an autoreleased instance. See MemoryManagement pages for info +( id ) pointInSpaceTimeWithX:( double ) x Y:( double )y Z:(double )z andT:(double) t; +( id ) pointInSpaceTime;

//default init method -( id ) initWithX:( double ) x Y:( double )y Z:(double )z andT:(double) t;

-( void ) setX:( double ) x; -( void ) setY:( double ) y; -( void ) setZ:( double ) z; -( void ) setT:( double ) t;

-(BOOL) isInPerfectHarmony; @end

@implementation PointInSpaceTime

+( id ) pointInSpaceTimeWithX:( double ) local_x Y:( double )local_y Z:(double )z andT:(double) local_t { return [ [ [ self alloc ] initWithX:local_x Y:local_y Z:local_z andT:local_t ] autorelease ]; }

+( id ) pointInSpaceTime { return [ [ [ self alloc ] init ] autorelease ]; }

-( id ) initWithX:( double ) local_x Y:( double )local_y Z:(double )z andT:(double) local_t { x = local_x; y = local_y; z = local_z; t = local_t; }

-( void ) setX:( double ) local_x { x = local_x; }

-( void ) setY:( double ) local_y { y = local_y; }

-( void ) setZ:( double ) local_z { z = local_z; }

-( void ) setT:( double ) local_t { t = local_t; }

-(BOOL) isInPerfectHarmony { if ((x==y) && (y==z) && (z==t) return TRUE; else return FALSE; }

@end

See the MemoryManagement page for discussion of the autorelease call you see above. You might use this class like this:

PointInSpaceTime *myPoint = [ PointInSpaceTime pointInSpaceTimeWithX:PI Y:MOLE Z:EARTH_GRAVITATIONAL_CONSTANT T:AGE_OF_DIRT_IN_EONS ];

if( [ myPoint isInPerfectHarmony ] ) { [ self attainZen ]; }

If you have other code dependent on the structure as it currently exists, you can wrapper it in an object:

@interface PointInSpaceTimeObject { @private PointInSpaceTime myPointInSpaceTime; }

//add methods here for getting and ( possibly ) setting the structure; @end

And yes, NSData can wrapper basic C memory chunks. See the docs. All are possible. And I’m sure several others I haven’t thought of. Was this just a curiousity question, or are you facing a particular problem? – TimHart

Actually, more of a design question. I want to develop a geometry class for use in SQLite with computational geometry methods. I have done some preliminary work using geometry objects, but it is just too slow for large database files. Structs are so much faster that I want to use them when possible, but also maintain the ease of ObjectiveC when possible. I thought of using them much the same as NSPoint, but with the possibility of 2 and 3 dimensions (maybe 4 later) and double precision. So right now I am thinking:

Geometry inherits from NSObject -NSString * geometryText2d [example: point(x y) ; line(x y, x y, x y) ; polygon((x y, x y, x y, x y)) ; polygonWithInnerRing((x y, x y, x y, x y)(x y, x y, x y, x y)) -basic methods

Point inherits from Geometry -PointInSpaceTime -Example Methods: parsePointFromGeometryText, pointInPolygon

Line inherits from Geometry -CArray (PointInSpaceTime) [Maybe an NSArray with PointInSpaceTime wrapped into NSData?] -Example Methods: parseLineFromGeometryText, clipToPolygon

Polygon inherits from Geometry -NSArray (Line) -Example Methods: parsePolygonFromGeometryText, intersection

Later MultiPoint MultiLine MultiPolygon

This will, I hope, speed up access to data, but maintain object oriented design benefits. I’ve had classes in C, but not C++ and am wary of starting down that road. All the C++ code I have seem looks excrutiatingly complicated for my experience level. However, looking at the code below, if I wrap the point up and never deal with it directly, I think I can manage if I can keep the Point, LIne, and Polygon as ObjectiveC classes, and wrap the points in some sort of array for the Line class. I am comfortable in ObjectiveC and prefer to keep code in either basic C or ObjectiveC as much as possible. Any feedback/concerns/suggetions?

Philip D Riggs

One option that may give you the best of both worlds is ObjectiveCee’s @defs compiler directive:

http://developer.apple.com/documentation/Cocoa/Conceptual/ObjectiveC/3objc_language_overview/chapter_3_section_8.html Basically, you CAN have your cake and eat it too.

Don’t completely disregard the CeePlusPlus suggestion below, though. Apple’s compiler allows you to mix and match both languages - to a degree. Called ObjectiveCeePlusPlus, you get the zero overhead possibilities of CeePlusPlus where you need speed/efficiency at runtime, along with the dynamism of ObjectiveCee where you feel it’s valuable.

The basics of C++ aren’t too complicated to handle if you know C and ObjectiveCee. Given your requirements, the combination of the two may be just what you’re looking for. – TimHart —-

I would suggest using CeePlusPlus for this kind of thing. That gives you much less overhead in the syntax, e.g.:

struct PointInSpaceTime { double values[4];

PointInSpaceTime (double x, double y, double z, double t) { double v[] = { x, y, z, t }; std::copy(beginof(v), endof(v), beginof(values)); }

bool operator== (PointInSpaceTime const& rhs) const { return std::equal(beginof(values), endof(values), beginof(rhs.values)); }

bool operator< (PointInSpaceTime const& rhs) const { return std::lexicographical_compare( beginof(values), endof(values), beginof(rhs.values), endof(rhs.values) ); }

… };

Now you can do stuff like:

PointInSpaceTime p1(1.0, 2.0, 3.0, 4.0); PointInSpaceTime p2(2.0, 1.0, 3.0, 4.0);

cout « (p1 == p2 ? “equal” : “not equal”) « endl;

Or:

PointInSpaceTime points[] = { p1, p2 }; std::sort(beginof(points), endof(points));

And much more…

(see CocoaSTL for definiation of beginof/ endof functions)