Looking for ways to deduce selector classifications - smalltalk

I'm working on an a SelectorClassifierAssistant object that suggests possible classifications for a method.
So far I've implemented the following strategies:
local classifications of same selector (i.e., in the same hierarchy up and down)
other classifications of the same selector in other classes (sorted by frequency)
accessor detection (is the method a setter or a getter? ifTrue: [^#accessing])
known selector prefixes (#is..., #print..., #as..., etc.)
and, if all of the above fail (i.e., suggestions is empty):
categories of other messages sent in the method (only the ones understood in the method's class and superclasses)
Are there other known heuristics I could use?
---EDIT---
selectors only sent from the same class -> #private (#Carlos E. Ferro)
known selectors (#= #hash, etc. -> #comparing and so on)
empty categories (if a category is empty chances are that the user just added it for classifying the next selector.)

I think that in your last option, instead of looking at categories of messages sent in the method to classify, it would be better to look for categories of the senders of this message.
This is a heuristic that I often use.
The reason I found is that methods called from this one most likely are general services, but senders of this one consider it as a service and define an ¨area of interest¨ where this message is used.
---- EDIT ----
Another related heuristic, very simple is this: if all senders are local (this class or subclasses), then "private" is a good candidate to classify it. Methods without senders out of its class are often private services.

Sure. Here are a number of other well-known patterns:
instance creation
newStringBrick
^ GLMStringBrick new
default values
defaultDotsText
^ '...'
lazy initialization
items
^items ifNil: [ items := OrderedCollection new ]
And of course there are also lots of framework/application specific patterns. Sunit unit tests, seaside applications, magritte descriptions all have specific patterns. a very simple one is the magritte-description
descriptionArticle
<magritteDescription>
^QCToOneOptionRelationDescription new
accessor: #article;
label: 'Article';
options: self possibleArticles;
priority: 40;
classes: QCArticle allSubclasses;
yourself.
[edit] And then the heuristics: another possibly good suggestion is the one I used last, or if you see a short repeating cycle the one that fits.
And would it be possible to find a good category for extension methods directly?

Related

Method and parameter naming conforming to Swift API Design Guideline

It's not obvious to me from reading the current API design guideline, which of the following version is better.
class MediaLoader {}
class MediaRequest {}
let mediaLoader = MediaLoader()
let mediaRequest = MediaRequest()
// Option 1
mediaLoader.add(request: mediaRequest)
// Option 2
mediaLoader.add(mediaRequest: mediaRequest)
// Option 3
mediaLoader.addRequest(mediaRequest)
// Option 4
mediaLoader.add(mediaRequest)
Which of the above conforms to the current API design guideline the best?
The answer really depends on the purpose and semantics of MediaLoader. If MediaLoader is only a collection of mediaRequests, then .add(mediaRequest) is the way to go because it would flow grammatically and be meaningful in context.
On the other hand, if a mediaRequest is merely one of many different things contributing to its purpose, then .add() alone would not convey enough context to properly read the statement. For example, if you could also add display channels or filters, then merely saying .add(something) would not be clear enough. This is when you would use an extended name that describes the relationship. e.g. .addRequest(), addChannel(), addFilter().
But not .add(request:...), because, using a name for the first parameter is not the ideal way to distinguish between relationships. It should be used instead to clarify the method by which the addition will be performed or the way the request will be accessed. This will leave the "nameless" variant for the most frequent and straightforward use case. e.g. .add(fromTemplate:webRequesTemplate) or .addRequest(fromTemplate:webTemplate).

difference between unidirectional association and dependency

According to wikipedia
Dependency is a relationship that shows that an element, or set of elements, requires other model elements for their specification or implementation.[1] The element is dependent upon the independent element, called the supplier.
So is it not the same as unidirectional association?
Do we use dependency when an operation in one class uses object of the other class as its parameter?
How are unidirectional association and dependency different.
Any example would be very helpful
Dependency :
Indicates that a client element(of any kind, including classes,
packages, use cases, etc) has knowledge of another supplier element,
and a change in supplier can effect the client.
So "dependency" is very broad relationship.Suppose that if a class-object(client) has another class-object(supplier) as a member,if a class-object send a message to another class-object,if a class-object takes another class-object as an parameter from its methods, even if a class(client) is subclass of another class(supplier) there will be dependency since change from supplier will effect clients.
Technically all of those relationships can be shown by "Dependency" line. But some of above relationships already has special notations: such as for superclass-subclass relationship we have generalization relationship.No need to show also "dependency" line because if they have generalization relationship, they have dependency. And we have "association" relationship for class-object(client) who has another class-object as a member [attribute]. So also no need to show extra dependency line in this situation.
Actually "Dependency" is badly defined relationship for class diagrams. But it can be usefull for showing dependency in which UML has no special notation such as :
if you has another class-object(supplier) as a parameter in one of your class(client) methods
if you have dependency to global variables
when you call static methods on another classes.
local variables (which you think you have important dependency)
public class RepositoryManager
{
public UpdatePriceFor(ProductDescription description)
{
Time date = Clock::GetTime();
Money oldPrice =description.GetPrice();
...
}
private IList<Item> itemsList = new List<Item>();
}
So all "associations" are also shows "dependency".But "dependency" is
broad-general-weak relationship.As a rule if there is a special
relationship which is more specific-stronger than dependency
relationship than use it. And lastly use all your relationship
"economically". Show only important ones based on modeler-model reader
perspectives.
[ Source : Adapted from Craig Larman's Applying UML and Patterns book ]
Check Fowlers bliki for further information DependencyAndAssociation
Association means that the two associated entities are linked semantically. Dependency only declares that there is a... well, dependency of some sort. All associations are dependencies, while a dependency does not actually mean association. For example, class 'A' depends on class 'B' if it has a method that takes 'B' and passes it as argument to a function in another class. But if 'A' calls some method of class 'B', it should be modeled as association.
Disclaimer I have read the UML specification and also asked myself this question a number of times. I arrived at at the definition above, but I'm still not sure it is 100% correct.

Whats is methodLists attribute of the structure objc_class for?

Looking inside the runtime.h, I found the definition of the structure objc_class.
Among various members, We have this :-
struct objc_method_list **methodLists
We definitely need to know what all methods a class has,
But a list of methods should be fine, but why do we have "lists" ?
Why not just one list ?
Also, can anyone specify that, Are methods inherited from superclass part of that list or we get to them via superclass pointer that points to parent class's structure.
Here is my detail investigation into struct objc_method_list **methodLists : http://blog.csdn.net/jasonblog/article/details/7303618
And in short, methodLists stores SEL-IMP mapping of the instance methods by default. In this situation, it has only one list.
As the name 'methodLists' suggests, it can contain more than one list. If you add a category to a class, the runtime system will insert one more list into methodLists, which points to the method-list of the category.
I tried to answer this question several months ago, but at that time SO discard my answer due to network problem. Now I meet it again :)
The purpose is explained in objc-class.m, as linked by Georg:
cls->methodLists may be in one of three forms:
NULL: The class has no methods.
non-NULL, with CLS_NO_METHOD_ARRAY set: cls->methodLists points
to a single method list, which is the class's only method list.
non-NULL, with CLS_NO_METHOD_ARRAY clear: cls->methodLists points to
an array of method list pointers. The end of the array's block
is set to -1. If the actual number of method lists is smaller
than that, the rest of the array is NULL.
Attaching categories and adding and removing classes may change
the form of the class list. In addition, individual method lists
may be reallocated when fixed up.
Classes are initially read as #1 or #2. If a category is attached
or other methods added, the class is changed to #3. Once in form #3,
the class is never downgraded to #1 or #2, even if methods are removed.
Classes added with objc_addClass are initially either #1 or #3.
The short answer is therefore "because of categories." When a category is injected, rather than try to combine its method list with the one existing list, a new entry is simply added to methodLists, and set to the list coming from the category. This probably makes category injection faster, since it avoids (potential) large reallocations and copying.

Difference between object and instance

I know this sort of question has been asked before, but I still feel that the answer is too ambiguous for me (and, by extension, some/most beginners) to grasp.
I have been trying to teach myself broader concepts of programming than procedural and basic OOP. I understand the concrete concepts of OOP (you make a class that has data (members) and functions (methods) and then instantiate that class at run time to actually do stuff, that kind of thing).
I think I have a handle on what a class is (sort of a design blueprint for an instance to be created in its likeness at compile time). But if that's the case, what is an object? I also know that in prototype based languages, this can muck things up even more, but perhaps this is why there needs to be a clear distinction between object and instance in my mind.
Beyond that, I struggle with the concepts of "object" and "instance". A lot of resources that I read (including answers at SO) say that they are largely the same and that the difference is in semantics. Other people say that there is a true conceptual difference between the two.
Can the experts here at SO help a beginner have that "aha" moment to move forward in the world of OOP?
Note: this isn't homework, I don't go to school - however, I think it would help people that are looking for homework help.
A blueprint for a house design is like a class description. All the houses built from that blueprint are objects of that class. A given house is an instance.
The truth is that object oriented programming often creates confusion by creating a disconnect between the philosophical side of development and the actual mechanical workings of the computer. I'll try to contrast the two for you:
The basic concept of OOP is this: Class >> Object >> Instance.
The class = the blue print.
The Object is an actual thing that is built based on the 'blue print' (like the house).
An instance is a virtual copy (but not a real copy) of the object.
The more technical explanation of an 'instance' is that it is a 'memory reference' or a reference variable. This means that an 'instance' is a variable in memory that only has a memory address of an object in it. The object it addresses is the same object the instance is said to be 'an instance of'. If you have many instances of an object, you really just have many variables in difference places in your memory that all have the same exact memory address in it - which are all the address of the same exact object. You can't ever 'change' an instance, although it looks like you can in your code. What you really do when you 'change' an instance is you change the original object directly. Electronically, the processor goes through one extra place in memory (the reference variable/instance) before it changes the data of the original object.
The process is: processor >> memory location of instance >> memory location of original object.
Note that it doesn't matter which instance you use - the end result will always be the same. ALL the instances will continue to maintain the same exact information in their memory locations - the object's memory address - and only the object will change.
The relationship between class and object is a bit more confusing, although philosophically its the easiest to understand (blue print >> house). If the object is actual data that is held somewhere in memory, what is 'class'? It turns out that mechanically the object is an exact copy of the class. So the class is just another variable somewhere else in memory that holds the same exact information that the object does. Note the difference between the relationships:
Object is a copy of the class.
Instance is a variable that holds the memory address of the object.
You can also have multiple objects of the same class and then multiple instances of each of those objects. In these cases, each object's set of instances are equivalent in value, but the instances between objects are not. For example:
Let Class A
From Class A let Object1, Object2, and Object3.
//Object1 has the same exact value as object2 and object3, but are in different places in memory.
from Object1>> let obj1_Instance1, obj1_Instace2 , obj1_Instance3
//all of these instances are also equivalent in value and in different places in memory. Their values = Object1.MemoryAddress.
etc.
Things get messier when you start introducing types. Here's an example using types from c#:
//assume class Person exists
Person john = new Person();
Actually, this code is easier to analyze if you break it down into two parts:
Person john;
john = new Person();
In technical speak, the first line 'declares a variable of type Person. But what does that mean?? The general explanation is that I now have an empty variable that can only hold a Person object. But wait a minute - its an empty variable! There is nothing in that variables memory location. It turns out that 'types' are mechanically meaningless. Types were originally invented as a way to manage data and nothing else. Even when you declare primitive types such as int, str, chr (w/o initializing it), nothing happens within the computer. This weird syntactical aspect of programming is part of where people get the idea that classes are the blueprint of objects. OOP's have gotten even more confusing with types with delegate types, event handlers, etc. I would try not focus on them too much and just remember that they are all a misnomer. Nothing changes with the variable until its either becomes an object or is set to a memory address of an object.
The second line is also a bit confusing because it does two things at once:
The right side "new Person()" is evaluated first. It creates a new copy of the Person class - that is, it creates a new object.
The left side "john =", is then evaluated after that. It turns john into a reference variable giving it the memory address of the object that was just created on the right side of the same line.
If you want to become a good developer, its important to understand that no computer environment ever works based on philosophic ideals. Computers aren't even that logical - they're really just a big collection of wires that are glued together using basic boolean circuits (mostly NAND and OR).
The word Class comes from Classification (A Category into which something is put), Now we have all heard that a Class is like a Blueprint,but what does this exactly mean ? It means that the Class holds a Description of a particular Category ,(I would like to show the difference between Class , Object and Instance with example using Java and I would request the readers to visualise it like a Story while reading it , and if you are not familiar with java doesn't matter) So let us start with make a Category called HumanBeing , so the Java program will expressed it as follows
class HumanBeing{
/*We will slowly build this category*/
}
Now what attributes does a HumanBeing have in general Name,Age,Height,Weight for now let us limit our self to these four attributes, let us add it to our Category
class HumanBeing{
private String Name;
private int Age;
private float Height;
private float Weight;
/*We still need to add methods*/
}
Now every category has a behaviour for example category Dog has a behaviour to bark,fetch,roll etc... , Similarly our category HumanBeing can also have certain behaviour,for example when we ask our HumanBeing what is your name/age/weight/height? It should give us its name/age/weight/height, so in java we do it as follows
class HumanBeing{
private String Name;
private int Age;
private float Height;
private float Weight;
public HumanBeing(String Name,int Age,float Height,float Weight){
this.Name = Name;
this.Age = Age;
this.Height = Height;
this.Weight = Weight;
}
public String getName(){
return this.Name;
}
public int getAge(){
return this.age;
}
public float getHeight(){
return this.Height;
}
public float getWeight(){
return this.Weight;
}
}
Now we have added behaviour to our category HumanBeing,so we can ask for its name ,age ,height ,weight but whom will you ask these details from , because class HumanBeing is just a category , it is a blueprint for example an Architect makes a blueprint on a paper of the building he wants to build , now we cannot go on live in the blueprint(its description of the building) we can only live in the building once it is built
So here we need to make a humanbeing from our category which we have described above , so how do we do that in Java
class Birth{
public static void main(String [] args){
HumanBeing firstHuman = new HumanBeing("Adam",25,6.2,90);
}
}
Now in the above example we have created our first human being with name age height weight , so what exactly is happening in the above code? . We are Instantiating our category HumanBeing i.e An Object of our class is created
Note : Object and Instance are not Synonyms In some cases it seems like Object and Instance are Synonyms but they are not, I will give both cases
Case 1: Object and Instance seems to be Synonyms
Let me elaborate a bit , when we say HumanBeing firstHuman = new HumanBeing("Adam",25,6.2,90); An Object of our category is created on the heap memory and some address is allocated to it , and firstHuman holds a reference to that address, now this Object is An Object of HumanBeing and also An Instance of HumanBeing.
Here it seems like Objects and Instance are Synonyms,I will repeat myself they are not synonyms
Let Us Resume our Story , we have created our firstHuman , now we can ask his name,age,height,weight , this is how we do it in Java
class Birth{
public static void main(String [] args){
HumanBeing firstHuman = new HumanBeing("Adam",25,6.2,90);
System.out.println(firstHuman.getName());
System.out.println(firstHuman.getAge());
...
...
}
}
so we have first human being and lets move feather by give our first human being some qualification ,let's make him a Doctor , so we need a category called Doctor and give our Doctor some behaviour ,so in java we do as follows
class Doctor extends HumanBeing{
public Doctor(String Name,int Age,float Height,float Weight){
super(Name,Age,Height,Weight);
}
public void doOperation(){
/* Do some Operation*/
}
public void doConsultation(){
/* Do so Consultation*/
}
}
Here we have used the concept of Inheritance which is bringing some reusability in the code , Every Doctor will always be a HumanBeing first , so A Doctor will have Name,Age,Weight,Height which will be Inherited from HumanBeing instead of writing it again , note that we have just written a description of a doctor we have not yet created one , so let us create a Doctor in our class Birth
class Birth{
public static void main(String [] args){
Doctor firstDoctor = new Doctor("Strange",40,6,80);
.......
.......
/*Assume some method calls , use of behaviour*/
.......
.......
}
}
Case 2: Object and Instance are not Synonyms
In the above code we can visualise that we are Instantiating our category Doctor and bringing it to life i.e we are simply creating an Object of the category Doctor , As we already know Object are created on Heap Memory and firstDoctor holds a reference to that Object on the heap ;
This particular Object firstDoctor is as follows (please note firstDoctor holds a reference to the object , it is not the object itself)
firstDoctor is An Object of class Doctor And An Instance of A class Doctor
firstDoctor is Not An Object of class HumanBeing But An Instance of class HumanBeing
So a particular Object can be an instance to a particular class but it need not be an object of that given class
Conclusion:
An Object is said to be an Instance of a particular Category if it satisfies all the characteristic of that particular Category
Real world example will be as follows , we are first born as Humans so image us as Object of Human , now when we grow up we take up responsibilities and learn new skills and play different roles in life example Son, brother, a daughter, father ,mother now What are we actually?We can say that we are Objects of Human But Instances of Brother,daughter,...,etc
I hope this helps
Thank You
Objects are things in memory while instances are things that reference to them. In the above pic:
std(instance) -> Student Object (right)
std1(instance) -> Student Object (left)
std2(instance) -> Student Object (left)
std3(instance) -> no object (null)
An object is an instance of a class (for class based languages).
I think this is the simplest explanation I can come up with.
A class defines an object. You can go even further in many languages and say an interface defines common attributes and methods between objects.
An object is something that can represent something in the real world. When you want the object to actually represent something in the real world that object must be instantiated. Instantiation means you must define the characteristics (attributes) of this specific object, usually through a constructor.
Once you have defined these characteristics you now have an instance of an object.
Hope this clears things up.
"A class describes a set of objects called its instances." - The Xerox learning Research Group, "The Smalltalk-80 System", Byte Magazine Volume 06 Number 08, p39, 1981.
What is an Object ?
An object is an instance of a class. Object can best be understood by finding real world examples around you. You desk, your laptop, your car all are good real world examples of an object.
Real world object share two characteristics, they all have state and behaviour. Humans are also a good example of an object, We humans have state/attributes - name, height, weight and behavior - walk, run, talk, sleep, code :P.
What is a Class ?
A class is a blueprint or a template that describes the details of an object. These details are viz
name
attributes/state
operations/methods
class Car
{
int speed = 0;
int gear = 1;
void changeGear(int newGear)
{
gear = newGear;
}
void speedUp(int increment)
{
speed = speed + increment;
}
void applyBrakes(int decrement)
{
speed = speed - decrement;
}
}
Consider the above example, the fields speed and gear will represent the state of the object, and methods changeGear, speedUp and applyBrakes define the behaviour of the Car object with the outside world.
References:
What is an Object ?
What is a Class ?
I think that it is important to point out that there are generally two things. The blueprint and the copies. People tend to name these different things; classes, objects, instances are just some of the names that people use for them. The important thing is that there is the blueprint and copies of it - regardless of the names for them. If you already have the understanding for these two, just avoid the other things that are confusing you.
Lets compare apples to apples. We all know what an apple is. What it looks like. What it tastes like. That is a class. It is the definition of a thing. It is what we know about a thing.
Now go find an apple. That is an instance. We can see it. We can taste it. We can do things with it. It is what we have.
Class = What we know about something. A definition.
Object/Instance = Something that fits that definition that we have and can do things with.
In some cases, the term "object" may be used to describe an instance, but in other cases it's used to describe a reference to an instance. The term "instance" only refers to the actual instance.
For example, a List may be described as a collection of objects, but what it actually holds are references to object instances.
I have always liked the idea that equals the definition of a class as that of an "Abstract Data Type". That is, when you defined a class you're are defining a new type of "something", his data type representation, in terms of primitives and other "somethings", and his behavior in terms of functions and/or methods. (Sorry for the generality and formalism)
Whenever you defined a class you open a new possibility for defining certain entities with its properties and behavior, when you instantiate and/or create a particular object out of it you're actually materializing that possibility.
Sometimes the terms object and instances are interchangeable. Some OOP purists will maintain that everything is an object, I will not complain, but in the real OOP world, we developers use two concepts:
Class: Abstract Data Type sample from which you can derive other ADT and create objects.
Objects: Also called instances, represents particular examples of the data structures and functions represented by a given Abstract Data Type.
Object Oriented Programming is a system metaphor that helps you organize the knowledge your program needs to handle, in a way that will make it easier for you to develop your program. When you choose to program using OOP you pick up your OOP-Googles, and you decide that you will see the problem of the real world as many objects collaborating between themselves, by sending messages. Instead of seeing a Guy driving a Car you see a Guy sending a message to the car indicating what he wants the car to do. The car is a big object, and will respond to that message by sending a message to it's engine or it's wheel to be able to respond properly to what the Driver told him to do in the message, etc...
After you've created your system metaphor, and you are seeing all the reality as objects sending messages, you decide to put all the things your are seeing that are relevant to your problem domain in the PC. There you notice that there are a lot of Guys driving different cards, and it's senseless to program the behavior of each one of them separately because they all behave in the same way... So you can say two things:
All those guys behave in the same way, so I'll create a class called
Driver that will specify who all the Drivers in the world behave,
because they all behave in the same way. (And your are using class based OOP)
Or your could say Hey! The second Driver behaves in the same way as the first Driver, except he likes going a little faster. And the third Driver behaves in the same way as the first Driver, except he likes zigzagging when he drives. (And you use prototype based OOP).
Then you start putting in the computer the information of how all the Drivers behave (or how the first driver behave, and how the second and third differ from that one), and after a while you have your program, and you use the code to create three drivers that are the model you are using inside that PC to refeer to the drivers you saw in the real world. Those 3 drivers that you created inside the PC are instances of either the prototype ( actually the first one is the prototype, the first one might be the prototype himself depending on how you model things) or the class that you created.
The difference between instance and object is that object is the metaphor you use in the real world. You choose to see the guy and the car as objects (It would be incorrect to say that you see them as instances) collaborating between themselves. And then you use it as inspiration to create your code. The instance only exists in your program, after you've created the prototype or the class. The "objects" exist outside the PC because its the mapping you use to unite the real world with the program. It unites the Guy with the instance of Driver you created in the PC. So object and instance are extremely related, but they are not exactly the same (an instance is a "leg" of an object in the program, and the other "leg" is in the real world).
I guess the best answer has already been given away.
Classes are blueprints, and objects are buildings or examples of that blueprint did the trick for me as well.
Sometimes, I'd like to think that classes are templates (like in MS Word), while objects are the documents that use the template.
Extending one of the earlier given examples in this thread...
Consider a scenario - There is a requirement that 5 houses need to be built in a neighbourhood for residential purposes. All 5 houses share a common construction architecture.
The construction architecture is a class.
House is an object.
Each house with people staying in it is an instance.

Polymorphism with different method signatures

I have a group of classes (say for validation rules). Each one returns a true or false.
I use id and call a method signature for each one of the classes and get the results allowing me to dynamically create validation rules.
Worked great until...
I have a new class that takes an extra parameter to come up with its validation.
What is the best way to deal with this?
Modify every other classes method signature to take a parameter that they don't need?
Probably the most appropriate course of action is to abstract your parameter passing into an object that can have a variable profile of variables.
Of course, more simply, Objective-C does allow for a variable parameter list much like C:
void method(int a, ...) // in C
- (void) method:(id) firstObject, ... // in ObjC
Apple has Technical Q&A on the very subject.