Protocol oriented Programming (PoP)

This post is a part of my blogpost series while learning Swift and Programming paradigms. First part on Functional Programming can be found here.

Protocol oriented programming (POP) can be said as as – a way to adhere the “Program to interface not to implement” principle – which is one of the basic design principles of software development. Structure your code such that you define interfaces first then make the classes implement it. Define interfaces using protocols and then make classes implement them. POP is just another technical term for the interface programming in Swift programming language.

Yes! There are additional advantages of POP in swift using “Extensions” along with protocols in Swift.

Let us discuss this with a classic example – Shapes and its classes.

Any shape has area. Every shape needs to be drawn on some canvas. Typically an average programmer thinks:

  • Let’s have a base class named Shape.
  • Let’s derive all other concrete classes from Shape class.

The first and foremost disadvantage of this solution is all concrete class inherits everything from base class whether its needed or not. Along with that multiple inheritance is not possible and moreover its a bad way.

In a practical world as the product grows its code grows and there will be a lot of overhead built over a period of time.

Here comes Protocol Oriented Programming – creating interfaces

This is link to Xcode playground @ github – giving a typical example of POP. Please go through the code before proceeding to understand the advantages.

Some plus points of POP :

  • No overhead of base class.
  • Protocol extensions are a great way to give default implementation.
  • Classes can conform to multiple protocols also use their default implementations without overhead of inheritance.
  • Restrict access ONLY to specific functionalities of an object.
  • Protocols can be conformed by structures & enumerations in Swift. Class-Only protocols can also be defined [ protocol TheClassOnlyProtocol: class { … } ].
  • Protocols can be used as a type for variables, which is more dynamic than declaring variables with concrete classes.
  • Delegation works as it used to in Objective-C.
  • Protocols can be used in Collection types
  • A Protocol can inherit from multiple other protocols (multiple inheritance)

I am a newbie to Swift. This is my understanding of POP. Please do correct me or ask your queries on the topic. Let’s share, discuss and grow together.

Hope you found this post useful to some extent.

What is Functional Programming?

swift

For quite sometime now I am on a new programming language from Apple – Swift. Apart from the hype Swift got there are multiple programming paradigms that I got introduced to – Functional Programming (FP), Protocol oriented programming (PoP).

I was little curious as I knew what functions are and what protocols are used for especially in objective-C, and my mind asked me few questions –

  • What are Functional Programming and Protocol oriented programming?
  • Is swift a Functional Programming language or a Protocol oriented programming language?
  • Is it mandatory to learn in Functional Programming or Protocol oriented programming to code in Swift?
  • What is the right way?

Browsing through some websites and reading few books, this is what I understand. In this post I will be sharing my thoughts & understanding on FP and a subsequent post on PoP.

Q: What is Functional Programming? Answer: (In non-developer’s term) Functional Programming is a way of how you code your solution, it’s a philosophy, it’s not related to any specific programming language.

Technical Answer: Functional programming is a way of programming where focus is on designing functions and their interactions rather than on data unlike Object Oriented Programming(OOP) where focus in on designing classes around data.

A functional programming language supports and encourages programming without side-effects. FP is a way in which you code only pure functions with all objects used in the function are passed as input parameters and all results are defined as output/return objects. In here, Functions are always pure and deterministic.

Now, what are pure functions? Answer: Functions that returns same value for given set of input values at any given point in time. These functions never modify the input values, never access, consume or modify anything outside of its scope.

Technical Answer: The result cannot depend on any hidden objects/values/state that change over a period of time. The function, In the process of execution, cannot cause any semantically observable side effects, like mutation of mutable objects or state change of any other objects.

Following holds true for FP:

  • In FP the focus is on how you design functions rather than modeling the data.
  • In FP utmost preference is given to immutability. FP hates mutating data.
  • In FP results are always same for same values supplied.
  • Functions in FP does not change any values supplied to it, neither change values outside of scope. Input parameters are only consumed but not modified.
  • In FP like Swift and Scala results are most of the time return in containers/optionals. (Yes, I dug into Scala programming language concept to understand few concepts of FP. Optionals in Swift are ‘inspired’ from Monads from Scala 😀 )

Advantages of Functional Programming:

  1. You definitely know what functions do and what they return. No side-effects.
  2. You need not worry much about wrapping data into threads, mutex and locks.
  3. Helps in writing lots of Unit test cases, increasing the code coverage.
  4. Modules can be isolated and independent from each other.
  5. Crash free: Functions return values wrapped in containers [Swift, Scala]

Keeping in mind all of this 2 questions raise in mind:

  1. Can we adopt functional programming in Swift? Yes, Definitely!
  2. Can we build iOS Apps in Swift completely with FP philosophy? Partially Yes. Why? Given my experience with the MV-X (MVC, MVVM, etc.,) patterns used in iOS app development access object properties and mutating datasources somewhere in the code is a must and cannot be avoided. So it is up to a developer to use their creativity to design functions or some modules in an app which adhere completely to FP.

I am new to FP and still experimenting with it, trying to understand it still more deeply. I also thought of adding few code snippets in this post to clearly explain and convert Impure functions into Pure ones, but surely we can discuss this in comments. Let’s discuss in comments section about these interpretations and understandings.

Keep visiting this blog for more programming stuff. Next post on Protocol oriented Programming.

Referred links which helped me understand:

 

Design Patterns & Principles

This is a part of a blogpost series listing the design principles and pattern definitions.

The Strategy Pattern

defines a family of algorithms, encapsulates each one, and makes them interchangeable. Strategy lets algorithm vary independently from clients that use it.

Following are few principles adhered in this pattern

  • Identify the aspect of your application that vary and separate them from what stays same.
  • Program to interface not to an implementation
  • Favor composition over inheritance

The example code for strategy pattern in Swift is available at my github repo.

courtesy: Head First Design Patterns

Compiler directives @ objective-C 2.0

Here are the list of directives in Objective-C 2.0. Lemme know if anything is missing.

// directives used in class declaration
@interface @public @package @protected @private @property @end

// directives used in class implementation
@implementation @synthesize @dynamic @end

// creating a string object
@”string”

// for forward declaration
@class

// similar to function pointers
@selector()

// used for method declaration mainly for delegation purpose
@protocol @required @optional @end

// to assure thread safety
@synchronized

// handling memory issues (useful in ARC enabled apps)
@autoreleasepool

// throwing and handling exceptions
@throw @try @catch @finally

// type encoding (Apple docs)
@encode

// aliasing an existing class name to a new one (useful in refactoring)
@compatibility_alias

[ Examples of all these directives are on the way 😉 ]

[reader enjoy:keep_programming];

Rotate Scale Flip – NSImageView

This is a problem which I have fiddled a lot to solve. Sometimes NSImageView objects seems to behave strangely when you rotate them and try to reposition it inside its superview.

The Problem:
I had an NSImageView as subview of another NSView. There where slider controls to scale and rotate the subview as and when required. I also had horizontal and vertical flipping options. So when I played around with it, the subview started acting weirdly.

I had used [subview setFrameCenterRotation:angle]; to rotate the subview.
For flipping NSAffineTransform scaleXBy:yBy: was helpful. Scaling I had set the imageScaling flag,
so that I just have to set the frame and NSImageView takes care of it by itself.

Now why is this acting weirdly? Seems all fine. All proper apis have been used. After some brain wracking and a little help from OpenGL transformation application, I found out the solution.

If you have rotated a image view, before doing any other transformation rotate it back to zero angle and the apply the transforms and then rotate it back to required angle.

Code for flipping images:

- (void)flipImageVertically:(NSImageView*)imageView
{
NSAffineTransform *flipper = [NSAffineTransform transform];
NSSize dimensions = [imageView frame].size;

NSImage *aFrame = [imageView.image copy];

[aFrame lockFocus];
[flipper scaleXBy:1.0 yBy:-1.0];
[flipper set];

[aFrame drawAtPoint:NSMakePoint(0,-dimensions.height)
fromRect:NSMakeRect(0,0, dimensions.width, dimensions.height)
operation:NSCompositeCopy fraction:1.0];

[aFrame unlockFocus];

[imageView setImage:aFrame];

[aFrame release];

}

- (void)flipImageHorizontally:(NSImageView*)imageView
{
NSAffineTransform *flipper = [NSAffineTransform transform];
NSSize dimensions = [imageView frame].size;

NSImage *aFrame = [imageView.image copy];

[aFrame lockFocus];
[flipper scaleXBy:-1.0 yBy:1.0];
[flipper set];

[aFrame drawAtPoint:NSMakePoint(-dimensions.width,0)
fromRect:NSMakeRect(0,0, dimensions.width, dimensions.height)
operation:NSCompositeCopy fraction:1.0];

[aFrame unlockFocus];

[imageView setImage:aFrame];

[aFrame release];
}

Tint an image view

I was developing an editor kind of stuff, where there are multiple images placed in a view and lot of custom drawings happen here and there in the custom views. The user gets to select a view and it has to be highlighted. For this I used to dram a boundary line using NSBezierPath in the view’s drawRect: method. But this used to hide the edges and corners of the view. Users started to complain they cannot place the views pixel perfect. They demanded for tinting the view.

After googling out for some time I found following NSImage extension:

@implementation NSImage (Extends)

- (NSImage *)imageTintedWithColor:(NSColor *)tint
{
if (tint != nil) {
NSSize size = [self size];
NSRect bounds = { NSZeroPoint, size };
NSImage *tintedImage = [[NSImage alloc] initWithSize:size];

[tintedImage lockFocus];

CIFilter *colorGenerator = [CIFilter filterWithName:@"CIConstantColorGenerator"];
CIColor *color = [[[CIColor alloc] initWithColor:tint] autorelease];

[colorGenerator setValue:color forKey:@"inputColor"];

CIFilter *monochromeFilter = [CIFilter filterWithName:@"CIColorMonochrome"];
CIImage *baseImage = [CIImage imageWithData:[self TIFFRepresentation]];

[monochromeFilter setValue:baseImage forKey:@"inputImage"];
[monochromeFilter setValue:[CIColor colorWithRed:0.75 green:0.75 blue:0.75] forKey:@"inputColor"];
[monochromeFilter setValue:[NSNumber numberWithFloat:1.0] forKey:@"inputIntensity"];

CIFilter *compositingFilter = [CIFilter filterWithName:@"CIMultiplyCompositing"];

[compositingFilter setValue:[colorGenerator valueForKey:@"outputImage"] forKey:@"inputImage"];
[compositingFilter setValue:[monochromeFilter valueForKey:@"outputImage"] forKey:@"inputBackgroundImage"];

CIImage *outputImage = [compositingFilter valueForKey:@"outputImage"];

[outputImage drawAtPoint:NSZeroPoint
fromRect:bounds
operation:NSCompositeCopy
fraction:1.0];

[tintedImage unlockFocus];

return [tintedImage autorelease];
}
else {
return [[self copy] autorelease];
}
}

@end

Call this method on the view’s image:
[view.image imageTintedWithColor:[NSColor redColor]];

Hope this is helpful.

Hello world ();

Hi (welcome)
{

This is a programming blog;
Here I planning to share my programming tips and tricks which I learn;

// Learning game programming now
Basically I share Mac / iPhone app and game development;
And currently want to learn development on Androids;

So I hope this new journey will be exciting.

thanks;

}