Talk:Object-oriented programming

I would like to rename this to object theory. I understand object-oriented programming is the most common term and nearly no one uses a word like object theory instead of it. But the trouble with the current title is OOP is a POV'd term. The situation is similar to that in socialism or communism. The majority of people claim class, inheritance and encapslation are foundamental aspects of OOP but some argue inheritance is not always necessary, for instance. It doesn't matter who is right. The fact is there are disputes regarding the definition of object-oriented programming. So it would be troublesome with the current name as some claim inheritance is an aspect of class-based OOP then the article sholdn't cover, while some say inheritance is a foundamental of OOP and there is no reason that the article doesn't cover it. I don't think such debate would be settled. The benefit of the title object theory is it is more abstract term. It can discuss languages before object-oriented programming. Ada8 something supports objects but is usually considered not OOP. Then where we should discuss such? A separate article? The title object theory implies it is about programming aspect regarding objects. It is a broader term that can contain class-based OOP, object-based programming and so on. Subprogram is an good example. People hardly uses a term subprogram over subroutines or functions. But in encyclopedia, use of such an abstract term is preferable because if the article named function, it implies a subprogram should be a function like in math, which is not the case. If named subroutine, it tends to emphasizes a behavior as side-effect and it makes hard to talk about function in functional languages. OOP is a similar case.

And as usual, if I don't see an object, I will go ahead. -- Taku 13:44 23 May 2003 (UTC)

If you want a separate article called object theory that explains differnet methods of using objects in programming languages then fine. But the aspects of the OO paradigm discussed in this article (abstraction, encapsulation, polymorphism and inheritance) is a vast subject that deserves a least a whole page devoted to it. What you have us call it? OOP is the universally understood term used for these concepts. Mintguy 14:01 23 May 2003 (UTC)

Please don't use the current content as an excuse. I want to rename first because otherwise, people will claim it is not OOP. Certainly we want to discuss what is OOP? But the more important is, history of OOP, point of views, variants, and so on. They are certainly beyond an article named object-oriented programming. Again, oftentimes, broader term serves well in encyclopedia. I showed an example above. -- Taku 15:11 23 May 2003 (UTC)

I happen to disagree with your choice of the word subprogram, in all my 25 years experience of being a programmer I can't recall anyone ever using the term subprogram for subroutine, function or procedure or method. However those four words that that you've grouped under subprogram do have different meanings to different people, usually depending on what language you are talking about. i.e. C programmers talk about functions when they mean procedures. I personally think subroutine is a better word because it is one that is in common usage and I beleive it has a longer history than the other words. But that is another issue.

Why don't we try to resolve this issue sensibly. Why don't you start writng an article about object theory and we'll see needs to be merged in from OOP? Mintguy 15:21 23 May 2003 (UTC)

About subprogram. That is exactly my point. There are a lot of concepts related to subroutine, function, procedure and methods. But we don't want to have a separate article for each one. What I am trying to do here is the same thing. Because for programmers, the difference is a huge matter, but for general audience, they want to know more about what is happening in cs. We don't want to teach them OOP should be bah bah bah. That is a difference seen in articles of wikipedia and textbooks. Because most of textbooks or websites have a POV such as OOP should be. Yeah, I can start certainly I can start an article called object theory but the trouble is we need to make sure history is combined. What about renaming this first, add more about object-based OOP and difference between class-based and object-based and then decide what name should be given or without renaming, start adding such point. This is just a chiken and egg problem. I don't care which one is first. -- Taku 16:17 23 May 2003 (UTC)

But why choose a word that no-one uses. That makes no sense whatsoever. I wouldn't expect to find an article about fishing under fish hunting. To the best of my knowledge, few people if any use the word subprogram to mean a subroutine or procedure. Subroutine is a word that assembler programmers generally use and it makes no assumption about whether there is a return value of side effect, or whether it is a member of a class or not. It just means a group of instructions that you jump to perform a task and then jump back from continuing execution where you left off.

I am against the idea of moving this page, because by adding stuff about other systems with objects or if you like widgets that do not embody OO principles you are diluting it and moving away from OOP. Mintguy 16:30 23 May 2003 (UTC)

See this is why I need to rename first. Anyway, we always make a redirect so of course, you can find out the article, object theory. If I don't mean to stick only to that term, but other possibilities are:

object theory
object-orientation
object-oriented
object-oriented paradigm

I just want to make the title more abstract, so that the article can contain parts beyond your strict definition of object-oriented programming. -- Taku 16:45 23 May 2003 (UTC)

Taku I am not going to argue with you anymore can we just leave this until we get some arbitration. Mintguy 16:55 23 May 2003 (UTC)

You mean mailing-list? Sure. By the way, I rewrote the subprogram so that it is more apparent that why the article is named as subprogram. Sometimes, we need to avoid common terms because they are too POV'd. -- Taku 17:04 23 May 2003 (UTC)

(NB I wrote the following before having read the latest round of your comments, Mintguy, Taku, but had an 'edit conflict')

I've just read and understood your arguments, Taku, but have to say I agree with Mintguy. Moving the content of this page to title that very few people use is not a good thing to do. With this in mind, am I correct in saying that you see two problems?:

There are aspects of programming (e.g inheritance) that some people call OOP and some people do not. i.e. there are different POVs about what OOP is.
There are uses of objects in programming (e.g. the objects in Ada) that no-one calls OOP.

Quick note: I presume you mean the old Ada83, which didn't have inheritance. The usual term for that was "object-based" programming, and it generally was NOT accepted as object-oriented programming. However, the current version of Ada (since 1995) has inheritance, and complete support for OOP. (Note added by User:dwheeler)

Yes, sorry, Taku writes above about "Ada 8 something". I should have written that too. Pcb21 18:08 23 May 2003 (UTC)

My view is that the best solution is that issue 1) can be discussed quite happily in this article. 2) requires a separate article e.g object (programming) or maybe object theory (not sure how widespread this term is and so am hesistant to agree to use it all), which links to this one, and devolves significant chunks of work to this article. That article could have more historical information than this article. Naturally this article would also link back to that one ... 'Programming languages may support objects but are rarely described as object-oriented languages. See object_(programming) for a more general article'. What do you think? Pcb21 17:16 23 May 2003 (UTC)

object theory would be a good additional place to discuss fine points of theory, but this article should be the general explanation of the range of what is called "object-oriented programming". It's not that big of a deal to say that there is a disagreement; NPOV means reporting points of view without trying to anoint any particular opinion as the "truth". ABC's assertion that "XYZ is not true object-oriented programming" just means that you report it as an assertion of ABC; removing the report entirely is taking the POV that ABC is so wrong that the assertion should be censored from the article. People that think they know the definition of object-oriented programming should probably excuse themselves from touching this article; a bald list of all the multiple definitions that have been used will make a longish article all by itself. (BTW, I have heard people use "subprogram" and even used it myself a couple times, but it's a somewhat archaic usage from the heyday of Fortran and Cobol - yes, I was there, but just a teenager I swear :-) ). Stan 19:13 23 May 2003 (UTC)

I was thinking and I agree that it's possible to discuss object-oriented programming in general, if not easy. There is a term that is quite popular and it should be strange if wikipedia doesn't cover that term mainly not as part of some article. But still I also think we should cover use of object in general, which can be part of OOP but can be outside. Besides, I realized it's possible to merge the article latter if needed. So I will write an article object theory I proposed. We will see and we can discuss again we should combine them or rename or anything else. -- Taku 21:45 25 May 2003 (UTC)

I disagree that "abstraction" is a standard definition of OOP nor does OOP have a monopoly on abstraction. For example, functions provide black boxing of implementation, and relational algebra is sometimes regarded as being highly abstract. The main text makes it appear that "abstraction" belongs with the "classic three". It does not.

OOP Criticism

I think this sentence has no bearing on OOP and should be removed:

Indeed, the rise of GUIs changed the user focus from the sequential instructions of text-based interfaces to the more dynamic manipulation of tangible components.

Discussion of classes doesn't belong in the first paragraph. Plenty of OO languages have objects but no classes (eg. self, cecil). -- P3d0

Taku, my friend, I appreciate your intent, but I think your latest changes to this page have only made it worse. I'm fighting the urge to back them out, and instead I'll think about it for a while and see if I can combine the best of the old and new versions...

Most of all, I don't claim to be an expert, but I have been in the field of OO programming for some time now; I have read on the topic fairly extensively; I have participated in (and even moderated for a time) the comp.object newsgroup; and I have never heard the term "Object Theory" used by actual practitioners of the OO programming paradigm. An encyclopedia is the place to document terminology, not to invent it, so I think all references to "Object Theory" ought to be deleted unless you can provide a reference to some external definition of the term. -- P3d0 01:29, 3 Oct 2003 (UTC)

Because I just made up the term. I just couldn't come up with good termiologies. You may know or not, there is a heated dispute about what is OOP between me and Mitguy. You are right; the article is too much inclined to particular kind of OOP. Classes, for example, are not main parts of OOP, so I just went editing. Let me know what parts you disagree with or just go ahead state your view. I know you have good knowledge and I will certainly apprecite your contributions. -- Taku

I've created these three class diagrams to show how Aggregation, Composition and Inheritance appears in UML.

Taku,

Stop reverting to old versions that are incorrect and pushing your own uniformed POV.

OOP is most certainly a paradigm not a style. Abstraction is a fundemental property of OOP. You should not confuse your own or others lack of understand of OO as meaning that OO is vague.

AIH I AM an expert in OO. I hold a BSc(Hons) in Computing Science, 15 years development expertise, including 10 in OO Software.

I think you are new to here presumablly. The definition of OOP is always disputed if you know the history of this article. OOP with use of classes is called class-based OOP and OOP without classes is called object-based OOP. It is a paradigm if you put that in the context of software development, but a style if you compare that with other programming sytles.

Abstraction is tricky question. In some sense, any programming style promotes abstraction. I think reusability and encapsulation is more specific and unique to OOP. I will show my points more in the article.

Besides, some people with good education and good professional experience understand nothing about OOP, though I don't know if you are the case or not. Please show us evidence of your POV not who you are. I hope you can be accustomed to the wikistyle. -- Taku 18:00, Oct 21, 2003 (UTC)

This argument is obnoxious. Worse, there is a distinct shortage of evidence for and against different approaches. Who you are and what you know is irrelevant to whether the information is generally correct. If an issue is debatable, leave it in talk or put it on your user page and link your user page here.

I suggest that it makes no sense to call OOP a "style" of programming. If something is part of a language, it's use can hardly be a style, on par with indenting and variable naming conventions. Paradigm is an abused word, but it is much more suitable to describe a feature of a language which is built in and common to a whole group of different languages with different but related approaches.

I would request that anyone of sufficient knowledge consider expanding on the distinction between OOP and standard procedural languages, especially since many OOP languages, notably the two most popular (Java and C++) are still procedural. Many discussions of OOP seem to suggest that objects have a sort of independent existence, as if they were autonomous entities like processes. The idea, for example, that they "communicate", while nice in abstraction, is really not correct, at least not in any way especially different from how functions talk to one another. Objects are not agents, after all, they are special data structures. Brent Gulanowski 20:05, 24 Oct 2003 (UTC)

You have a point. My apology for stating OOP as style. Now I think the intro and the definition section is good enough, it is time to rework more following details. -- Taku

Taku,

The dispute between us is not about the definition of OOP. I think we agree well enough on that paradigm and the elements section described in this article. My disagreement is with you saying that a structured programming design (the Windows API) was originally designed as an OOP design. The two are different concepts. OOP includes structured programming as one of its component concepts but that doesn't make all structured design OOP design. JamesDay 20:48, 25 Oct 2003 (UTC)

Hope I'm not entering a hot-bed or starting a fire-storm. I made changes to the basic definition of OOP. I've studied OOP (in college and outside) and taught OOP (to corporate and government clients) over the past 20 years, and the consensus I have always found was OOP means: 1.) use of objects (duh; this often goes unstated), 2.) abstraction (yes, its used elsewhere but OOP requires), 3.) encapsulation, 4.) polymorphism (I added this; wasn't there; how could it be forgotten? OOP absolutely requires this), and 5.) inheritance (yes object-based can omit inheritance, but OOP requires.)

I dropped "reusability" because it is more of a "state of mind" than a tenet of OOP. Yes OOP encourages reusability, but my reusable component may not be sufficient for someone else, so reusability is something you approach, not something you are. We can look at the other items (abs,enc,poly,inh) and say definitively (in most cases; I'm sure some wise-a** can find an obscured example to the contrary) that a given language either does or does not have those attributes.

So, I hope my thinking is not off base and you like my edits (enough to keep them :)

MikeSchinkel 02:19, 26 Oct 2003 (UTC)

No one mind your contribution. An attempt is more hopeful than giving up. I think four points you pointed out are quite accepted--that's what I have learned and have seen in many books of OOP. My intial motivation to omit them is I wanted to turn the article into more general term rather than common one. The explanation of polymorphism is a good instance. Not every OOP language has a data type. I mean when you teach OOP, it is inevitable to explain about porymophism but it is actually rather an implementation choice and detail. Anyway, this is my reason. -- Taku 02:39, Oct 26, 2003 (UTC)

Thanks. Can you give me examples of OOP languages that don't have data types? Or are you meaning ones don't differentiate between integer, boolean, etc? My use of data type here, by the way, was specifically to avoid the concept of "class" because a class is a type of data type though not vice-versa, and some languages don't support class in the traditional sense and I didn't want to have to go into the intense detail required to explain. A good definition should be consise, no? :) MikeSchinkel 02:51, 26 Oct 2003 (UTC)

I think OOP without any data type can be possible, or of course it depends on the definition of OOP though. -- Taku 04:39, Oct 29, 2003 (UTC)

Can someone provide examples of "object-based" languages? Are we talking languages like JavaScript/JScript? Can someone fill out more what "object-based" means? TIA. MikeSchinkel 02:58, 26 Oct 2003 (UTC)

I think so, JavaScript, Visual Basic and such. I know we usually classify them as non-OOP. -- Taku 04:39, Oct 29, 2003 (UTC)

Ahh! I read your reply and thought "Of course! What was I thinking?" Then I went back and read the entry on Object-oriented_programming and remembered what I was thinking. The confusion I had was the distinction between "class-based" and "object-based" languages. I had never heard it explained that way so I was confused. Now I find it not only confusing but also pretty much inaccurate. The entry describes "Class-based" as "In this model, objects are entities that combine both state (i.e., data) and behavior (i.e., procedures, or methods)." That applies to "object-based" as well. I would say the inclusion of the term "class-based" should be removed, and then a reference to "object-based" be left in that describes it as "object-oriented minus inheritance."

There is another spin that could be added to this; that of OOP languages with staticly defined classes (i.e. most of them) and then ones with dynamically defined classes (i.e. javascript and a few others I'm sure.) However, I don't have enough experience with javascript to elaborate correctly.

Thoughts? MikeSchinkel 10:54, 2 Nov 2003 (UTC)

Somewhere I got the idea that there now existed one or more examples of object-oriented functional languages, but come to think of it, is that possible? Hmm, I guess it is if the objects never change, meaning no side effects, but what is an object if it isn't an abstract data type? How do you have ADTs without primitive data types?

... well, google helped: http://www.pasteur.fr/~letondal/object-based.html (Object-based PLs). OScheme would be a good example, and its interesting because it's untyped, I think. I guess that's the functional one I was thinking of. Brent Gulanowski 18:53, 26 Oct 2003 (UTC) Brent Gulanowski 18:53, 26 Oct 2003 (UTC)

If we must talk about OOP modelling the "real world" then can we please do it less prominently? OOP applies quite well in situations where there is no "real world" (eg. nodes in a control-flow graph) so I don't consider real-world fidelity to be very important at all. However, some disagree, so if someone feels like putting that back in, I won't remove it again. --P3d0 13:37, 3 Jan 2004 (UTC)

I think the description on abstraction is misleading. How would that abstraction be different from composition? Composition is a way of grouping multiple entitites into a single entity. Abstraction should be thought more as ignoring irrelevant details. --esap