> The article describes how the Linux kernel, despite being written in C, embraces object-oriented principles by using function pointers in structures to achieve polymorphism.
This technique predates object oriented programming. It is called an abstract data type or data abstraction. A key difference between data abstraction and object oriented programming is that you can leave functions unimplemented in your abstract data type while OOP requires that the functions always be implemented.
The sanest way to have optional functions in object oriented programming that occurs to me would be to have an additional class for each optional function and inherit each one you implement alongside your base class via multiple inheritance. Then you would need to check at runtime whether the object is an instance of the additional class before using an optional function. With an abstract data type, you would just be do a simple NULL check to see if the function pointer is present before using it.
> In Smalltalk and Objective-C, you just check at runtime whether an object instance responds to a message. This is the original OOP way.
This introduces performance issues larger than the typical ones associated with vtable lookups. Not all domains can afford this today and even fewer in the 80s/90s when these languages were first designed.
> It's sad that OOP was corrupted by the excessively class-centric C++ and Java design patterns.
Both Smalltalk and Objective-C are class based and messages are single-receiver dispatched. So it’s not classes that you’re objecting to. It’s compile-time resolved (eg: vtable) method dispatch vs a more dynamic dispatch with messages.
Ruby, Python, and Javascript all allow for last-resort attribute/message dispatching in various ways: Ruby via `method_missing`, Python by supplying `__getattr__`, and Javascript via Proxy objects.
When you fear about the branch miss, then you can also just call the method and catch the SIGSEGV. I think if you allow for the possibility of there being no implementation, then you can't really not have some decision there. This would also apply for say a C++ virtual method.
Are you talking about C? In C it's not guaranteed whether a non existing function will result in an actual function call. As soon as an actual function call was generated by the compiler, a modern CPU is very likely to trap.
We are talking about an optimization of a language implementation here. This would be very much written in a ASM or another language were this is defined.
Actually I would say that it is sad that developers learn a specific way how a technology is done in language XYZ and then use it as template everywhere else, what happened to the curiosity of learning?
The curiosity of learning is infeasible given that there are >15,000 programming languages. You might say to only learn the major/influential languages, but I have found my curiosity to wane with every additional language that I learn. So far, I have either used or dabbled in C, C++, FORTRAN, SML/NJ, PHP, Java, JavaScript, Python, Go, POSIX Shell, Assembly and SQL stored procedures. That is not to mention different dialects/versions of those, and the fact that assembly itself is a catch all category in which I have dabbled in multiple languages too. I am also not including languages for which I spent a minuscule amount of time (say a few hours), such as D (as in DTrace), Objective-C, Perl, COBOL and LISP. Then there is the misadventure into C# I took before I actually understood programming where I got stuck and dropped it. I am also not sure if I should mention AWK, as I only use it to select the Nth field in a bunch of lines of text when doing shell scripting and nothing else. Thus, I have used it many times, yet know almost none of its syntax.
Every few years, someone tells me I should learn another language, and in recent years, there just is no desire in my mind to want to learn yet another language that is merely another way of doing something that I already can do elsewhere and the only way I will is if I am forced (that is how I used Go).
That said, I do see what you are saying. C++ for example has an “support every paradigm” philosophy, so whenever someone who has learned C++ encounters a language using a paradigm that C++ assimilated, there is a huge temptation to try to view it through the lens of C++. I also can see the other side too: “C++ took me forever to learn. Why go through that again when I can use C++ as a shortcut to understand something else?”. C++ is essentially the Borg of programming languages.
I did not know they say the same thing about programming languages.
Based on his comment, I did not think that he is proficient in them, but that he has used them, which is fair enough, so have I, sans all the ones tied to either Apple (Swift) or Microsoft (C#).
I have some projects in Haskell just for curiosity's sake, and because what I wanted seemed like it would be nice in Haskell, and it indeed looks quite elegant to me, for this one particular project. Haskell is not a language I would use generally. OCaml is.
My proficiency in them varies. I have used them for projects, such that I can write code in them (provided I have references to read), but when I do, I only learn/use the subset of the language that I need. If I want to read code others have written, then I need to know the subset that they used and that is not always the subset that I know. Most of them are languages that I have used for a handful of projects or in the case of assembly, for very small portions of projects, with at least half of that spend just trying to understand how good of a job the compiler did in critical loops.
There are some languages in which I am extremely proficient. My best language is C, which is my favorite and I have used most features of every version of C from C89 to C11. My second best is probably either C++ or POSIX shell (although I have moments where I forget certain syntax and need to look it up, especially in POSIX shell for variants on variable substitution, e.g. ${VAR%%foobar}). I have used most features of C++98 and some from newer versions. My experiences with C++ have soured me on it, so I now try to avoid C++ whenever I can in favor of C and Python.
My first language was actually PHP 4.2.y, and I was fairly proficient in it, having spent a long time learning it while simultaneously writing my own code for a website as a teenager. However, I never once touched the portions describing objects/classes, namespaces or exceptions. Someone else at work writes Modern PHP code using Symfony and I have taken a peek at it. It looks very different from the PHP I knew because it uses the features I had avoided learning (and probably some new language features too), although I can sort of read it thanks to having learned those concepts in other languages.
I used SML/NJ and Java in college. Years after college, I modified an open source Android TV application written in Java to add some things I wanted, although honestly, beyond that I have not really touched either language. Give me an arbitrary application written in either to improve and I would have some difficulty, although I will probably be able to do it after filling the gaps in my understanding (and doing plenty of head banging if it is a large program).
I have used JavaScript for a few recent projects via electron/nodejs and I have done several small things in Python over the past several years. Each time, I only worked with the subset that I need. I am far from a master of either language that can understand arbitrary code written in them, but I am able to manage as far as my specific needs are concerned.
I could continue listing my experiences doing things in languages (like the time in college that I wrote some basic programs in FORTRAN 90 to try to learn it), but they really are not that interesting. It was often a project here or a small application there, and as I readily admitted, I only used a subset of most of the languages. For programming, a subset of commonly used bits is often all you need.
This is a very good point. A greater distinction needs to be made about class-based versus class-free OOP. Many of the problems or constraints around OOP that people don't like, come back to classes.
Often class-based programming is confused as being the only style of OOP, superior to all other styles, or heavy-handedly pushed on others. Many programmers are perfectly fine with using objects or only specific features of OOP, without classes, if they are "allowed" to.
Because Objective-C being based on C, means it would never be safe while remaining compatible with C.
Also dynamic runtime dispatch Smalltalk style can never be as fast as the VMT based dispatch, or compile time dispatch via generics, even with all the optimizations in place, that objc_msgSend() has had during its lifetime.
Still, Metal is implemented in Objective-C, so there is that.
There is always a subset of the population for which it is fashionable to try to make a new language to make things easier. It is inevitable that someone in a sufficiently large organization will try to make one, provided management supports it. Apple is a very wealthy company that is happy to sponsor R&D, had someone interested in this that they sponsored and the result was appealing enough that they decided to adopt it.
That is my understanding of how the process generally works and what I am willing to guess happened. Prior to this, they had been making incremental changes to Objective-C.
That said, from what I have seen of the syntax of both languages, swift’s syntax is nicer and that is not something that they would have been able to get from Objective-C. They already had tried syntax reform for
Objective-C once in the past and abandoned it.
I largely agree, and use these patterns in C, but you’re neglecting the usual approach of having a default or stub implementation in the base for classic OOP. There’s also the option of using interfaces in more modern OOP or concept-style languages where you can cast to an interface type to only require the subset of the API you actually need to call. Go is a good example of this, in fact doing the lookup at runtime from effectively a table of function pointers like this.
My point is that this pattern is not object oriented programming. As for a default behavior with it, you usually would do that by either always adding the default pointer when creating the structure or calling the default whenever the pointer is NULL.
In the Linux VFS for example, there are optimized functions for reading and writing, but if those are not implemented, a fallback to unoptimized functions is done at the call sites. Both sets are function pointers and you only need to implement one if I recall correctly.
To be fair, OOP is not 100% absolutely perfectly defined. Strustrup swears C++ is OOP, Alan Key, at least at some point laughed at C++, and people using CLOS have yet another definition
> My point is that this pattern is not object oriented programming.
I think the "is/is not" question is not so clear. If you think of "is" as a whether there's a homomorphism, then it makes sense to say that it is OOP, but it can qualify as being something else too, ie. it's not an exclusionary relationaship.
Object oriented programming implies certain contracts that the compiler enforces that are not enforced with data abstraction. Given that object oriented programming and data abstraction two live side by side in C++, we can spot the differences between member functions that have contracts enforced, and members function pointers that do not. Member functions have an implicit this pointer, and in a derived class, can call the base class version via a shorthand notation to the compiler (BaseClass::func() or super()), unless that base class version is a pure virtual function. Member function pointers have no implicit this pointer unless one is explicitly passed. They have no ability to access a base class variant via some shorthand notation to the compiler because the compiler has no contract saying that OOP is being done and there is a base class version of this function. Finally, classes with unimplemented member functions may not be instantiated as objects, while classes with unimplemented member functions pointers may.
If you think of the differences as being OOP implies contracts with the compiler and data abstraction does not (beyond a simple structure saying where the members are in memory), it becomes easier to see the two as different things.
So you can opt out or in to syntactic sugar, that makes C++ an interesting and useful language, but how you implement OOP, doesn't really affect if it is OOP.
> Object oriented programming implies certain contracts that the compiler enforces
Sorry, but where did you got this definition from? I've always thought OOP as a way of organizing your data and your code, sometimes supported by language-specific constructs, but not necessarily.
Can you organize your data into lists, trees, and hashmaps even if your language does not have those as native types? So you can think in a OO way even if the language has no notion of objects, methods, etc.
Member function pointers and member functions in C++ are two different things. Member function pointers are not OOP. They are data abstraction.
The entire point of OOP is to make contracts with the compiler that forcibly tie certain things together that are not tied together with data abstraction. Member functions are subject to inheritance and polymorphism. Member function pointers are not. Changing the type of your class will never magically change the contents of a member function pointer, but it will change the constants of a non-virtual member function. A member function will have a this pointer to refer to the class. A member function pointer does not unless you explicitly add one (named something other than this in C++).
OOP cannot be a formalization of what predated it because the predating patterns support things that OOP explicitly disallows, like instantiation with unimplemented functions. That is extremely useful when you want to implement an optional function, or mutually exclusive functions such that you pick which is optional. This should be the case in the Linux VFS with ->read() and ->read_iter(). Also, ASTs were formalized after OOP, despite existing prior to it in Lisp.
For full disclosure, I have never verified that leaving ->read() unimplemented when ->read_iter() is implemented is safe, but I have seen enough examples of code that I strongly suspect it is and if it is not, it is probably a bug.
OOP only disallows inheritance with unimplemented functions when it's a contract violation.
So that is to say, if the base class has a certain function which must be implemented and must provide certain behaviors, then the derived class must implement that function and provide all those behaviors.
The POSIX functions like read and write do not have a contract which says that all implementations of them must successfully transfer data. Being unimplemented (e.g returning -1 with errno EOPNOTSUPP or whatever) is allowed in the contract.
OOP just wants a derived thing to obey the contract of the abtraction it is inheriting, so if you want certain liberties, you have to push them into the contract.
I would call returning something being implemented as a stub rather than being unimplemented. When something is unimplemented and you try to call it, you crash due to a NULL/invalid pointer dereference, not get an error back. Of course, as far as getting things done is concerned, the two might as well be the same, but as far as how the language works, the two are different.
I think this is more of an effect of C distinguishing between allocating memory (aka object creation) and initialization, which other languages disallow for other reasons, not because there are not OOPy enough.
The concept of abstract data type is a real idea in the days of compiler design. You might as well say "compiler design predates object oriented programming". The technique described in the lead is used to implement object-oriented programming structures, just as it says. So are lots of compiler design features under the hood.
source- I wrote a windowing framework for MacOS using this pattern and others, in C with MetroWerks at the time.
You caused me to do some digging. That publication is dated November 1962. The Lisp 1.5 manual’s preface is dated August 17, 1962, which is even older. It describes lambdas and property lists, which seem like they can be used to implement ADTs, although I do not have a Lisp 1.5 interpreter since those are obsolete, so I cannot verify that. Computer history articles claim that Simula, the first object oriented language, was born in May 1962, but was not actually operational until January 1965:
Thus, while I had thought Lisp had ADT concepts before the first OOL existed, now I am not sure. My remark that they originated in Lisp had been said with the intention that I was talking about the first language to have it. The idea that the concept had been described outside of an actual language is tangential to what I had intended to say, which is news to me. Thanks for the link.
and the paper even starts with a critique of the efficiency of Lisp's approach for representing data with cons pairs (citing McCarthy's paper from the same year).
You might also want to watch Casey's great talk on the history of OOP
You can do exactly what was done in C with most OOP languages like Java & C# because you have lambdas now, and lambdas are just function pointers. You can literally assign them to instance variables (or static variables).
(sorry it took more than a decade for Java to catch up and Sun Microsystems originally sued Microsoft for trying to add lambdas to java way back when, and even wrote a white paper insisting that anonymous inner classes are a perfectly good substitute - stop laughing)
The interesting thing is, that in the OOP implementation inheritance IS composition of vtables and data. It's really only syntactic sugar, that is sometimes not unambiguous.
This is not quite correct. OOP implementation inheritance involves a kind of "open recursion" (that is, calls to base-class methods can end up dispatching to implementations from a derived class) that is not replicated with pure composition. All method calls, including method calls that originate from code in some class anywhere in the hierarchy, ultimately dispatch through the vtable of whatever object they're called on.
But that's exactly the same you would need to implement manually when you use composition. When constructing, you also need to construct the contained objects, when doing something that should affect a contained object, you need to dispatch it.
When a method is never overridden, it doesn't need to be in the vtable.
But you can do it. Actually you can call instance methods of other classes and change the class of an instance in Python like in C, but this dynamism is probably what makes it slow. Also doing that will make the program quite complicated, more so than in C, since Python also abstracts about this.
An abstract data type is a software design pattern.
The difference is that design patterns are a technique where you use features not implemented by the compiler or language, and all the checks have to be done by the developer, manually.
Thus, you are doing part of the work of the compiler.
In assembler, a function call is a design pattern.
For people wondering what it looks like without the syntactic sugar of carbon then look here [0]. As far as I can see, there's no support for parametric polymorphism.
> Having to pass the object explicitly every time feels clunky, especially compared to C++ where this is implicit.
I personally don't like implicit this. You are very much passing a this instance around, as opposed to a class method. Also explicit this eliminates the problem, that you don't know if the variable is an instance variable or a global/from somewhere else.
Agreed, one of the biggest design mistakes in the OOP syntax of C++ (and Java, for that matter) was not making `this` mandatory when referring to instance members.
C++ and Java went for the "objects as static closures" route, where it doesn't make any sense to have a `this`. Or, they made them superficially look like static closures, which in hindsight was probably not the best idea. Anyway, Java lets you use explicit `this`, I don't recall whether C++ makes it into a footgun or not.
Both languages let you use explicit `this` but don’t mandate it. The “static closure” approach is great. I don’t like having to explicitly pass `this` as a parameter to every method call as in the OP (or worse, the confusing Python approach of forcing `self` to be explicitly written in every non-static method signature but having it be implicitly passed during method calls).
What I don’t like is being able to reference instance members without `this`, e.g.
void foo() {
int x = bar + 1; // should be illegal: it can be hard to distinguish if `bar` is a local variable versus an instance member
int y = this->bar + 1; // disambiguation is good
}
> int x = bar + 1; // should be illegal: it can be hard to distinguish if `bar` is a local variable versus an instance member
If it was this->bar it could be a member, it could also be a static variable. A bar on its own could be local or it could be in any of the enclosing scopes or namespaces. Forcing "this" to be explicit doesn't make the code any clearer on its own.
The guy was referring to the explicit case where bar is a member variable. The cases where it is in the local scope under scoping rules or the global scope are not really an issue, since you can check the function definition to find if it is local. In the case that it is not in the function definition, then it is in the global scope in C. If implicit this were not done in C++, that would also be the case in C++, provided you do the sane thing and use the std namespace for everything. Just thinking about the headaches namespaces could cause when looking for such definitions with cscope gives me yet another reason to stay away from C++ whenever possible.
that's not really true - unlike a regular pointer, `this` is not allowed to be null, thus removing `if(this == nullptr)` is always a valid optimization to do.
Only in so much that this being null is UB, but in the real world it very much can be null and programs will sometimes crash deep inside methods called on nullptr.
Supporting an explicit this is required to be able to access the member variable when a local variable hides it under scoping rules. As another person replied, C++ does indeed have an implicit this.
Where not only is it explicit, but you need to specify the object twice (once to resolve the Vtable, and a second time to pass the object to the stateless C method implementation).
It would be an improvement on C++ since you would avoid the horrendous error messages from ridiculously long types, slow compile times, exception handling nightmares, bloat from RTTI and constant worry that operators might not mean what you expect. That is before even mentioning entirely new classes of problems like the diamond problem. Less is more.
That said, similar macro magic is used in C generic data structures and it works very well.
What guarantee do you have that ->ops is a vtable? It could contain function pointers that don’t take an implicit this argument like struct file_operations in Linux does. It could also contain variables that are non-pointers. Neither is allowed in a vtable, but both are fine to do in C.
You can take the API for a linked list and implement a balanced binary search tree behind it, but continuing to call it a linked list after doing that is wrong. Similarly, you can do pointer indirections the same way a vtable would do them, but if the things you get are not the equivalent of member function pointers, it is not a vtable.
> In computer programming, a virtual method table (VMT), virtual function table, virtual call table, dispatch table, vtable, or vftable is a mechanism used in a programming language to support dynamic dispatch (or run-time method binding). (Wikipedia)
When its a table of function pointers used for dynamic dispatch, to me it's a vtable. I don't care about their type signatures as long as they logically belong to the object in question.
You seem to have a different very narrow definition of vtables, so the discussion is kind of useless.
Yes I know. From the caller that might seem to be redundant, my argument was about the callee's side. Also it is not truely redundant, as you can write:
I think both of these invocations are invalid. Using object1's vtable methods on object2, obviously, but in the latter case: the vtable method should just point at the superclass impl, if not overridden. And if overridden and the child impl needs to call the superclass, it can just do so without dispatching through some vtable.
When you think of vtables as unique or owned by an object, then these example seem weird to you. When you think of them as orthogonal to your types/objects, these examples can be useful.
In the first example, object1 and object2 can very much be of the same type or compatible types/subtypes/supertypes. Having vtables per object as opposed to per class to me indicates, that it IS intended to modify the behaviour of an object by changing it's vtable. Using the behaviour of another object of the same type to treat the second object, seams valid to me.
In the second case, it's not about the child implementation dispatching to the superclass, it's about some external code wanting to treat it as an object of the supertype. It's what in other languages needs an upcast. And the supertype might also have dynamic behaviour, otherwise you of course wouldn't use a vtable.
I think it is wrong/weird for objects of the same type to have different vtables, yes. I would call those different types.
Upcasting is fine, but generally speaking the expected behavior of invoking a superclass method on an object that is actually a subclass is that the subclass method implementation is used (in C++, this would be a virtual/override type method, as opposed to a static method). Invoking a superclass-specific method impl on a subclass object is kind of weird.
That is a close match for the first example, with additional arguments.
It is helpful to remember that this is not object oriented programming and not try to shoehorn this into the paradigm of object oriented programming. This is data abstraction, which has similarities (and inspired OOP), but is subtly different. Data abstraction does not automatically imply any sort of inheritance. Thus you cannot treat things as necessarily having a subclass and superclass. If you must think of it in OOP terms, imagine that your superclass is an abstract class, with no implemented members, except you can instantiate a child class that is also abstract, and you will never do any inheritance on the so called child class.
Now, it is possible to implement things in such a way where they actually do have something that resembles a subclass and a superclass. This is often done in filesystem inode structures. The filesystem will have its own specialized inode structure where the generic VFS inode structure is the first member and thus you can cast safely from the generic inode structure to the specialized one. There is no need to cast in the other direction since you can access all of the generic inode structure’s members. This trick is useful when the VFS calls us via inode operations. We know that the inode pointer is really a pointer to our specialized inode structure, so we can safely cast to it to access the specialized fields. This is essentially `superclass->op->start(object)`, which was the second example.
Data abstraction is a really powerful technique and honestly, object oriented programming rarely does anything that makes me want it over data abstraction. The only thing that I have seen object oriented programming do better in practice than data abstraction is marketing. The second example is similar to C++’s curiously recurring template pattern, which adds boilerplate and fighting with the compiler with absurdly long error messages due to absurdly verbose types to achieve a result that often at best is the same thing. On top of those headaches, all of the language complexity makes the compile times slow. Only marketing could convince someone that the C++ OOP way is better.
I don't agree that your example pattern matches to the example I'm complaining about. vfs_rename() is using old_dir's vtable on old_dir. The vtable matches the object.
> Also explicit this eliminates the problem, that you don't know if the variable is an instance variable or a global/from somewhere else.
People typically use some kind of naming convention for their member variables, e.g. mFoo, m_Foo, m_foo, foo_, etc., so that's not an issue. I find `foo_` much more concise than `this->foo`. Also note that you can use explicity this in C++ if you really want to.
In code I write, I can know what variables mean. The feature loses its point, when it's not mandatory. Also being explicit allows you to be more expressive with variable name and ordering.
I don't quite agree, especially because the implicit this not only saves you from explicitly typing it, but also because by having actual methods you don't need to add the struct suffix to every function.
My problem with implicit this is more, that you can access member variables, without it being explicit, i.e. about the callee, not about the caller.
For the function naming, nothing stops you from doing the same in C:
static dosmth (struct * s);
s->dosmth = dosmth;
That doesn't stop you from mentioning s twice. While it is redundant in the common case, it isn't in every case like I wrote elsewhere. Also this is easily fixable as written several times here, by a macro, or by using the type directly.
This is not the same, you introduced dynamic function resolution (i.e.a function pointer tied to a specific instance), we are talking about static function resolution (purely based on the declared type).
“this” is a reserved keyword in C++, so you do not need to worry about it being a global variable.
That said, I like having a this pointer explicitly passed as it is in C with ADTs. The functions that do not need a this pointer never accidentally have it passed from the developer forgetting to mark the function static or not wanting to rewrite all of the function accesses to use the :: operator.
If you see "i++" in code and you don't have any context about what "i" is, then what difference does it make if "i" is a member variable, global variable, parameter, etc etc...
If all you see in code is a very tiny 3 character expression, you won't be able to make much of a judgement about it to begin with.
Not allowing a variable to implicitly refer to a member variable makes it much easier to find. If it is not declared in the function and there is no implicit dereferencing of a this pointer, the variable is global. If the variable name is commonly used and it is a member variable, it is a nightmare to hunt for the correct declaration in the codebase with cscope.
Good point. I had misunderstood the previous comment as suggesting that this be passed to the member function as an explicit argument, rather than requiring dereferences of this be explicit. The latter makes far more sense and I agree it makes reasoning about things much easier.
Hi, I don't know much about this. But it seems to me that the OP is doing it differently than the kernel devs. If you read the article that the OP links, then you get the impression that the vtables contain typed function pointers, while OP uses void pointers. Also the main benefit mentioned in the kernel dev article is that you save memory, by not having multiple function pointers in each structure instance, but instead you have just one pointer to a vtable in each instance. Thus the main benefit is saving memory according to kernel dev, but OP uses this vtable as a form of indirection to implement runtime method swapping and polymorphism, which is not even mentioned in the kernel dev article. Thus, OP uses some other pattern than the one mentioned by kernel dev.
OP doesn't use void pointers, he uses void. He writes about functions having no arguments and returning nothing for the same reason other blog posts name functions foo and bar.
> OP uses this vtable as a form of indirection to implement runtime method swapping and polymorphism
The kernel uses vtables to implement polymorphism, it doesn't store the vtable in the object to save space. If there is no polymorphism, you don't use a vtable at all, that's saving even more space.
> If there is no polymorphism, you don't use a vtable at all, that's saving even more space.
Not true. You use a vtable even if there is no polymorphism, in cases where you want to have objects store pointers to their methods (OO interface), but don't want each instance to have pointers to all methods. I was referring to this article, which the OP links in his post https://lwn.net/Articles/444910/
When do objects need to have function pointers? If the function never changes per object, then you would just directly call the function. When do you use function pointers? When the caller only knows the type of struct foo_ops, but not the actual called function. That's polymorphism.
You can redefine the methods at runtime. If for example I have a logger that logs on the local drive and another one that logs on a networked drive I can swap the method pointer. This is not polymorphism since the object does have to change. I can also load the function from a shared library that is unknown during compilation and use that as part of the objects methods.
I would call that polymorphism, since the object type the caller sees, still stays the same, i.e. the vtable itself or some containing struct/class. So as far as the caller is concerned, the object he has access to behaves differently.
I've done this on a few smaller projects when I was in college. It's fun bringing something similar to OOP into C; however you can get into trouble really quickly if you are not careful.
Note that this is using interfaces (i.e. vtables, records of function pointers), not full object-orientation. Other OO features, like classes and inheritance, have much more baggage, and are often not worth the associated pain.
In exact that same scenario, that you would cast to a subclass in another language, it's about language support for what for example the kernel does with container_of.
Of course casting to a subclass isn't guaranteed to succeed always, but for example when you have actually declared it as the subclass elsewhere it's fine without checking for isinstance.
vtables contain function pointers to functions that take “this” pointers. The author mentions struct file_operations as an example of a vtable. struct file_operations contains a pointer to a function that does not take “this” pointer. It is not even a vtable.
I would still call it a vtable. Who assures you that every function of a class needs a pointer to the object instance? When you don't need it, you can just leave it of when you roll your own.
Static member functions are tied to the class while virtual member functions are tied to the object. If you throw the static member functions into a vtable, you will at best have a bug where the static member function from a different class can be called. Alternatively, you would have an undefined behavior where the other class in the hierarchy does not implement this function.
There is a saying “If Your Only Tool Is a Hammer Then Every Problem Looks Like a Nail”. That is precisely what is happening here with the insistence to call what appears to be all structures of function pointers vtables. A vtable is something that follows a fairly well defined pattern for implementing inheritance. Not all things containing function pointers are vtables.
> Static member functions are tied to the class while virtual member functions are tied to the object.
That's nice, but the entire point is, that the caller doesn't know the type of the object, it only has a supertype. That's why you need dynamic dispatch here. Of course you can implement dynamic dispatch without function pointers, but it is done with function pointers here. If you don't want to name dynamic dispatch implemented with function pointers a vtable, OK, that's fine, but that's the definition I am familiar with.
The vtable pointer corresponds to the actual type in languages that use vtables to implement inheritance, so you do know the type.
Read my previous comment for the bug that would happen if what us being used in Linux were actually used for dynamic dispatch when implementing inheritance, and it should be clear this is something similar, but different.
Yup. I've often wonder why the aversion to C++ since they are obviously using objects. Is it that they don't want to also enable all the C++ language junk like templates or OO junk like inheritance?
Here's one example. For us, it's more a tradeoff rather than an aversion. There's pros (manual memory management in C) and cons (manual memory management in C) for each. We do math operations (dense and sparse matrix math for setting up and solving massive systems of differential equations) on massive graphs with up to billions of nodes and edges. We use C in parts of the engine because we need to manage memory at a very fine level to meet performance demands on our tool. Other parts of the tool use C++ because they decided the tradeoff benefited in the other direction, re memory access / management / ease of use. As a result we need really robust qa around memory leaks etc. and tbh we rely on one generational talent of an engineer to keep things from falling apart; but we get that speed. As a side note, we implement objects in C a little more complex than the op, so that the object really does end up as a black box to the user (other engineers), with all the beauty of data agnosticism.
VLAs, named structure assignment, sane treatment of the void type, having different "lifetimes" for object (the C understanding of object) existence and initialization, having different namespaces for composed types and variables, a local error handling convention, leading to better error messages, robuster behaviour and a feeling for completeness, a lot of examples of the article and here in the thread, and most importantly no magic.
If you're talking about conversion rules, explicit cast from void* to T* makes more sense than implicit since it is a downcast.
> VLAs
IMO these are a mistake (and C++ templates remove a lot of the need for it). They give a false sense of security and invite stack boundary overrun as many people forget to check bounds on these. I found and reported an unauthenticated RCE DoS (crash) in a distributed DB due to VLAs; worse, one cannot always assume the minimum stack size on a system.
> a local error handling convention
Exceptions are problematic in their implementation and how they are (mis)used, but they are supposed to be orthogonal to normal control flow handling, and are not supposed to replace it. They are more-or-less recoverable panics
C makes it obvious were you use that dynamism and where you don't. Syntactic sugar doesn't really make that much of a difference and also restricts more creative uses.
The C syntax is not really that complicated. Dynamic dispatch and virtual methods was already in the article. Here is inheritance:
struct Subclass {
struct Baseclass base;
};
That's not really that complicated. Sure, you need to encapsulate every method of the parent class, if you want to expose it. But you are also recommended to do that in other languages, and if you subclass you probably want to slightly modify behaviour anyway.
As for stuff like templates: C doesn't thinks everything needs to be in the compiler. For example shadowing and hiding symbols can be done by the linker, since this is the component that handles symbol resolution across different units anyway. When you want templates, either you actually want a cheap way of runtime dynamism, then do that, or you want source code generation. Why does the compiler need to do that? For the basics there is a separate tool in the language: the Preprocessor, if you want more, you are free to choose your tool. If you want a macro language, there is e.g. M4. If you want another generator just use it. If you feel no tool really cuts it, why don't you write your code generator in C?
I always wonder, why not anything similar made it into a new (some) C version? Clearly, there is a significant demand for - lots of people reimplementing the same (similar) set of patterns.
Whenever you invent syntactic sugar you need to make some usage blessed and some usage impossible/needing to fallback to the old way without syntactic sugar. See https://news.ycombinator.com/item?id=45040662. Also some point of C is, that it doesn't hide that dynamic complexity. You always see when there is dynamic dispatch. There are tons of language, which introduce some formalism for these concepts, honestly most modern imperative languages seem to be. The unique selling point of C is, that you see the complexity. That influences you to only use it if you really want it. Also the syntax isn't really that complicated.
I was involved in a product with a large codebase structured like this and it was a maintainability nightmare with no upsides.
Multiple attempts were made to move away from this to no avail.
Consider that the code has terrible readability due to no syntax-sugar, the compiler cannot see through the pointers to optimise anything, tooling has no clue what to do with it.
On top of that, the syntax is odd and requires any newbies to effectively understand how a c++ compiler works under-the-hood to get anything out of it.
On top of those points, the dubious benefits of OOP make doing this a quick way to kill long-term maintainability of your project.
For the devs who come after you, dont try to turn C into a poor-mans C++.
If you really want to, please just use C++.
Can you elaborate what exactly the maintainability nightmare was?
To me less syntactic sugar is more readable, because you see what function call involves dynamic dispatch and which doesn't. Ideally it should also lead to dynamic dispatch being restricted to where it is needed.
I don't know where (might also have been LWN), but there was a post about it actually being more optimizable by the compiler, because dynamic code in C involves much less function pointers and the compiler can assume UB more often, because the assignments are in user code.
> requires any newbies to effectively understand how a c++ compiler
You are not supposed to reimplement a C++ compiler exactly, you are supposed to understand how OOP works and then this emerges naturally.
> dont try to turn C into a poor-mans C++
It's not poor-mans C++, when it's idiomatic C.
People like me very much choose C while having this usage in mind, because its clearer and I can sprinkle dynamism where it's needed not where the language/compiler prescribes it and because every dynamism is clear because there is not dynamic sugar, so you can't hide it.
Another cool thing about this approach is you can have the arguments to your object init be a pointer to a structure of args. Then down the line you can add features to your object without having to change all the calls to init your object throughout the code base.
If this is the pattern you prefer, why not choose a language that caters to it? Choosing C just seems like you're TRYING to shoot yourself. I don't care how good you are at coding, this is just a bad decision.
> Ok so... why choose C if they know they're shooting themselves?
> Because they like how C caters to this.
We(aka I) think we are shooting ourselves less, because C represents the algorithms more in a way how we want to express them. C's lack of syntactic sugar means dynamic dispatch is always visible. C not prescribing which function pointers you can use, means that the most fitting way can be chosen as described by the article and the LWN post, as opposed to shoehorning it into some paradigm prescribed by the language, which causes more problems done the line.
This technique predates object oriented programming. It is called an abstract data type or data abstraction. A key difference between data abstraction and object oriented programming is that you can leave functions unimplemented in your abstract data type while OOP requires that the functions always be implemented.
The sanest way to have optional functions in object oriented programming that occurs to me would be to have an additional class for each optional function and inherit each one you implement alongside your base class via multiple inheritance. Then you would need to check at runtime whether the object is an instance of the additional class before using an optional function. With an abstract data type, you would just be do a simple NULL check to see if the function pointer is present before using it.