Background
I have an abstract class, something like
class IConverter{
public:
virtual void DoConvertion() = 0;
};
There will be many concrete classes which just implements DoConvertion method.
class TextConverter : public IConverter{
public:
virtual void DoConvertion(){
// my code goes here
}
};
class ImageConverter : public IConverter{
public:
virtual void DoConvertion(){
// my code goes here
}
};
There will be many concrete implementation like this. I have created a header file say, CharacterConverter.h which has the abstract class IConverter.
Question
Since my concrete classes just implement the DoConvertion method, is it required to create separate header files for each concrete class? I mean is it required to create ImageConverter.h, TextConverter.h and so on for all concrete classes? All these header files is going to contain the same code like IConverter abstract class.
Any thoughts?
It is not required. It's basically a judgment call.
If the implementation is simple for each class you can put them all in one .h and one .cpp
If the implementations are a bit longer, then it's probably cleaner to use a separate .h and .cpp file for each.
Some advantages of using a different .h/.cpp for each class:
It will keep the code organized and clean
Reduced compiling work: A change in one of the implementations won't need to recompile all others
Faster compiling time: Several compilers can compile multiple files at once such as Visual Studio's /MP switch. With several files you'll have a faster compile time.
Other files can include only what they need instead of everything
Faster link time: Linking time will be reduced due to incremental linking
Using version control you can look back on only the changes to a particular derived class, instead of having to go through all changes made to the massive 1 .h/.cpp file to find that one change in a particular derived class.
One of the main points of creating an interface class is so that clients can be depend on the abstract interface rather than the concrete implementation, and you are then free to change the implementation without impacting clients.
Putting the concrete declarations in the same header files as the interface declarations defeats this, so now if you change an implementation detail of a concrete class, your clients would need to re-compile.
Something you might consider, depending on the rest of your design, is a factory, where your abstract class has a static method (or multiple static methods, depending on how you implement it) that constructs the appropriate subclass and returns it as an IConverter*. With this, you can expose only the abstract definition in the header file, and have all the concrete class definitions and implementations in a single .cpp file along with the super class implementation. This gets a bit unwieldy if your subclass are large, but with smaller classes it reduces the number of files you have to manage.
But, as others have pointed out, it's ultimately a judgment call. The only performance issues would be related to compiling; more cpp files might take (slightly) longer to compile and more header files might increase dependency analysis. But there's no requirement that every header file have a matching cpp and vice verse.
Based on the comments, I'd recommend a structure like this:
IConverter.h ==> definition of IConverter
Converters.h ==> definitions of all subclasses
IConverter.cpp ==> include IConverter.h and Converters.h, contain implementation of IConverter abstract functionality (static factory method and any inheritable functionality)
TextConvter.cpp, ImagerConverter.cpp, etc. ==> seperate cpp files for each subclass, each containing IConverter.h and Converters.h
This allows you to only include the IConverter.h in any clients that use the factory and generic functionality. Putting all the other definitions in a single header allows you to consolidate if they're all basically the same. Separate cpp files allow you to take advantage of the compiler benefits mentioned by Brian. You could inline the subclass definitions in header files as mentioned, but that doesn't really buy you anything. Your compiler is usually smarter than you are when it comes to optimizations like inline.
You'll probably get answers both ways.
I'd say, for any trivial converters, having all of them in a single .h/.cpp pair is sufficient and that it's overkill to split every one into a single pair. I think the tradeoff of maintenance of lots of files vs. maintenance of a bunch of methods within a single file is worth it in this case.
Complex conversions probably deserve their own file pairs.
You will need definitions of the concrete classes to create objects, so you'll need to put those definitions into a .h file somewhere. Which file you put them in is up to you.
The best answer to this is what's easier to read. One long source file is going to be difficult for you and other programmers to follow. On the other hand, many tiny (half screen-full) source files is just as bad.
You'd probably be better off using factories or function pointers.
However, one particularly nasty way that springs to mind is using a macro to declare your concrete classes. For example:
At the bottom of IConverter.h include the following macro
#define DECLARE_CONVERTER_CLASS(CLASS_NAME) \
class CLASS_NAME : public IConverter\
{ \
public: \
CLASS_NAME() {} \
virtual void DoConversion(); \
}; \
Then in MyConverter1.cpp
DECLARE_CONVERTER_CLASS(MyConverter1)
virtual void MyConverter1::DoConversion()
{
...
}
Yuck :-)
Related
This question already has answers here:
Is the PIMPL idiom really used in practice?
(12 answers)
Closed 8 years ago.
Backgrounder:
The PIMPL Idiom (Pointer to IMPLementation) is a technique for implementation hiding in which a public class wraps a structure or class that cannot be seen outside the library the public class is part of.
This hides internal implementation details and data from the user of the library.
When implementing this idiom why would you place the public methods on the pimpl class and not the public class since the public classes method implementations would be compiled into the library and the user only has the header file?
To illustrate, this code puts the Purr() implementation on the impl class and wraps it as well.
Why not implement Purr directly on the public class?
// header file:
class Cat {
private:
class CatImpl; // Not defined here
CatImpl *cat_; // Handle
public:
Cat(); // Constructor
~Cat(); // Destructor
// Other operations...
Purr();
};
// CPP file:
#include "cat.h"
class Cat::CatImpl {
Purr();
... // The actual implementation can be anything
};
Cat::Cat() {
cat_ = new CatImpl;
}
Cat::~Cat() {
delete cat_;
}
Cat::Purr(){ cat_->Purr(); }
CatImpl::Purr(){
printf("purrrrrr");
}
I think most people refer to this as the Handle Body idiom. See James Coplien's book Advanced C++ Programming Styles and Idioms. It's also known as the Cheshire Cat because of Lewis Caroll's character that fades away until only the grin remains.
The example code should be distributed across two sets of source files. Then only Cat.h is the file that is shipped with the product.
CatImpl.h is included by Cat.cpp and CatImpl.cpp contains the implementation for CatImpl::Purr(). This won't be visible to the public using your product.
Basically the idea is to hide as much as possible of the implementation from prying eyes.
This is most useful where you have a commercial product that is shipped as a series of libraries that are accessed via an API that the customer's code is compiled against and linked to.
We did this with the rewrite of IONA's Orbix 3.3 product in 2000.
As mentioned by others, using his technique completely decouples the implementation from the interface of the object. Then you won't have to recompile everything that uses Cat if you just want to change the implementation of Purr().
This technique is used in a methodology called design by contract.
Because you want Purr() to be able to use private members of CatImpl. Cat::Purr() would not be allowed such an access without a friend declaration.
Because you then don't mix responsibilities: one class implements, one class forwards.
For what is worth, it separates the implementation from the interface. This is usually not very important in small size projects. But, in large projects and libraries, it can be used to reduce the build times significantly.
Consider that the implementation of Cat may include many headers, may involve template meta-programming which takes time to compile on its own. Why should a user, who just wants to use the Cat have to include all that? Hence, all the necessary files are hidden using the pimpl idiom (hence the forward declaration of CatImpl), and using the interface does not force the user to include them.
I'm developing a library for nonlinear optimization (read "lots of nasty math"), which is implemented in templates, so most of the code is in headers. It takes about five minutes to compile (on a decent multi-core CPU), and just parsing the headers in an otherwise empty .cpp takes about a minute. So anyone using the library has to wait a couple of minutes every time they compile their code, which makes the development quite tedious. However, by hiding the implementation and the headers, one just includes a simple interface file, which compiles instantly.
It does not necessarily have anything to do with protecting the implementation from being copied by other companies - which wouldn't probably happen anyway, unless the inner workings of your algorithm can be guessed from the definitions of the member variables (if so, it is probably not very complicated and not worth protecting in the first place).
If your class uses the PIMPL idiom, you can avoid changing the header file on the public class.
This allows you to add/remove methods to the PIMPL class, without modifying the external class's header file. You can also add/remove #includes to the PIMPL too.
When you change the external class's header file, you have to recompile everything that #includes it (and if any of those are header files, you have to recompile everything that #includes them, and so on).
Typically, the only reference to a PIMPL class in the header for the owner class (Cat in this case) would be a forward declaration, as you have done here, because that can greatly reduce the dependencies.
For example, if your PIMPL class has ComplicatedClass as a member (and not just a pointer or reference to it) then you would need to have ComplicatedClass fully defined before its use. In practice, this means including file "ComplicatedClass.h" (which will also indirectly include anything ComplicatedClass depends on). This can lead to a single header fill pulling in lots and lots of stuff, which is bad for managing your dependencies (and your compile times).
When you use the PIMPL idiom, you only need to #include the stuff used in the public interface of your owner type (which would be Cat here). Which makes things better for people using your library, and means you don't need to worry about people depending on some internal part of your library - either by mistake, or because they want to do something you don't allow, so they #define private public before including your files.
If it's a simple class, there's usually isn't any reason to use a PIMPL, but for times when the types are quite big, it can be a big help (especially in avoiding long build times).
Well, I wouldn't use it. I have a better alternative:
File foo.h
class Foo {
public:
virtual ~Foo() { }
virtual void someMethod() = 0;
// This "replaces" the constructor
static Foo *create();
}
File foo.cpp
namespace {
class FooImpl: virtual public Foo {
public:
void someMethod() {
//....
}
};
}
Foo *Foo::create() {
return new FooImpl;
}
Does this pattern have a name?
As someone who is also a Python and Java programmer, I like this a lot more than the PIMPL idiom.
Placing the call to the impl->Purr inside the .cpp file means that in the future you could do something completely different without having to change the header file.
Maybe next year they discover a helper method they could have called instead and so they can change the code to call that directly and not use impl->Purr at all. (Yes, they could achieve the same thing by updating the actual impl::Purr method as well, but in that case you are stuck with an extra function call that achieves nothing but calling the next function in turn.)
It also means the header only has definitions and does not have any implementation which makes for a cleaner separation, which is the whole point of the idiom.
We use the PIMPL idiom in order to emulate aspect-oriented programming where pre, post and error aspects are called before and after the execution of a member function.
struct Omg{
void purr(){ cout<< "purr\n"; }
};
struct Lol{
Omg* omg;
/*...*/
void purr(){ try{ pre(); omg-> purr(); post(); }catch(...){ error(); } }
};
We also use a pointer-to-base class to share different aspects between many classes.
The drawback of this approach is that the library user has to take into account all the aspects that are going to be executed, but only sees his/her class. It requires browsing the documentation for any side effects.
I just implemented my first PIMPL class over the last couple of days. I used it to eliminate problems I was having, including file *winsock2.*h in Borland Builder. It seemed to be screwing up struct alignment and since I had socket things in the class private data, those problems were spreading to any .cpp file that included the header.
By using PIMPL, winsock2.h was included in only one .cpp file where I could put a lid on the problem and not worry that it would come back to bite me.
To answer the original question, the advantage I found in forwarding the calls to the PIMPL class was that the PIMPL class is the same as what your original class would have been before you pimpl'd it, plus your implementations aren't spread over two classes in some weird fashion. It's much clearer to implement the public members to simply forward to the PIMPL class.
Like Mr Nodet said, one class, one responsibility.
I don't know if this is a difference worth mentioning but...
Would it be possible to have the implementation in its own namespace and have a public wrapper / library namespace for the code the user sees:
catlib::Cat::Purr(){ cat_->Purr(); }
cat::Cat::Purr(){
printf("purrrrrr");
}
This way all library code can make use of the cat namespace and as the need to expose a class to the user arises a wrapper could be created in the catlib namespace.
I find it telling that, in spite of how well-known the PIMPL idiom is, I don't see it crop up very often in real life (e.g., in open source projects).
I often wonder if the "benefits" are overblown; yes, you can make some of your implementation details even more hidden, and yes, you can change your implementation without changing the header, but it's not obvious that these are big advantages in reality.
That is to say, it's not clear that there's any need for your implementation to be that well hidden, and perhaps it's quite rare that people really do change only the implementation; as soon as you need to add new methods, say, you need to change the header anyway.
If we have to declare proto type inside of public tag for class then what is the benefit of making functions outside of class
IMHO, there are mainly 2 reasons:
a matter of style:
Some of us prefere keeping individual blocks of code small. Putting the method definitions outside of the class definition helps in keeping the class definition small. You may or not agree on that.
separate compilation:
A common usage is to put the class definition in an include file to allow other compilation units to use it. As it will be included in several files, it will be compiled several times. So it can make sense on a performance build point of view to keep that include file small.
In addition, usages recommend to keep individual files in a projet small. If one is changed, only that one need to be compiled again and archived in a code version system. Splitting the class definition and its methods in 2 different files helps in that goal.
Anyway, both styles are allowed per the standard, and it has no impact on run time performance
For now I do not use header files at all. Classes are each in a single .cpp file completely. But to save compile time I want to make use of header files now. My hope is that Visual Studio won't compile classes which weren't modified for debug builds then.
Is there a way to mention only public methods and members in the header file. In theory that would be enough information for the compiler. If another file, say main.cpp includes the class header there is no need for private methods and members, is it?
How can I use header files without retyping the names of private methods and members? The reasons for me to want so is coding productivity. When I want do add a small helper function to the class used by another method, I don't want to have to also add it's signature to the header file.
If another file, say main.cpp includes the class header there is no need for private methods and members, is it?
No, public methods and members aren't necessarily enough. For example, if another .cpp file were to try and create an instance of your class:
SomeClass instance;
the compiler will need to know, among other things, how much memory to allocate for SomeClass. For that it requires full knowledge of SomeClass's private data members.
The way you are framing the question makes it sound as if you were intent on fighting the language. I don't think that's a good way to go about it. I think the best way is to do things the way things are usually done in the language of your choice, and depart from that only when there is a specific, clearly understood need.
The way things are usually done in C++ is that the entire class declaration goes in the header file, and the definition is in some way split between the header file and the corresponding .cpp file. The exact split is determined by various technical considerations. For example, templates and inline functions normally have to appear in the header file. On the other hand, placing code in header files increases dependencies and potentially build times.
There are ways to address these issues. However, since this involves extra complexity, I'd argue that this should only be done if there is a clearly identifiable need.
I don't know of a way to do what you're asking, but there is another way to give some isolation. You might want to take a look at the pimpl idiom as it offers isolation about private information. It's a little bit of extra work, but it can be extremely useful, especially in large projects.
All function declarations should go in header files and all function definitions should go in cpp files. It's not good coding practice to put declarations inside the cpp files.
You could put definitions inside headers though, when you write templates or inline functions.
Once you declare the class in the header file, you have to declare all its methods and members inside the class' declaration in the header, given that your class is no longer declared in the cpp file.
I know it's possible to do class implementation in more than one file(yes, I know that this is bad idea), but I want to know if it's possible to write class definition in separate files without getting a redefinition error (maybe some tricks, or else...)
No, not the same class, but why would you?
You could define two classes with the same name in the same namespace (the same signature! That will actually be the same class at all, just defined in two ways) in two different header files and compile your program, if your source files don't include both headers. Your application could even be linked, but then at runtime things won't work as you expect (it'll be undefined behavior!), unless the classes are identical.
See Charles Bailey's answer for a more formal explanation.
EDIT:
If what you want is to have a single class defined by more than one file, in order to (for example) add some automatically generated functions to your class, you could #include a secondary file in the middle of your class.
/* automatically generated file - autofile.gen.h */
void f1( ) { /* ... */ }
void f2( ) { /* ... */ }
void f3( ) { /* ... */ }
/* ... */
/* your header file with your class to be expanded */
class C
{
/* ... */
#include "autofile.gen.h"
/* ... */
};
Not as nice and clean as the partial keyword in C#, but yes, you can split your class into multiple header files:
class MyClass
{
#include "my_class_aspect1.h"
#include "my_class_aspect2.h"
#include "my_class_aspect3.h"
};
#include "my_class_aspect1.inl"
#include "my_class_aspect2.inl"
#include "my_class_aspect3.inl"
Well sort of ...
If you have 1 header file as follows
ClassDef1.h:
class ClassDef
{
protected:
// blah, etc.
public:
// more blah
and another as follows
ClassDef2.h:
public:
// Yet more blah.
};
The class will effectively have been defined across 2 files.
Beyond that sort of trickery .. AFAIK, no you can't.
Yes, you can. Each definition must occur in a separate translation unit but there are heavy restrictions on multiple definitions.
Each definition must consist of the same sequence of tokens and in each definition corresponding names must refer to the same entity (or an entity within the definition of the class itself).
See 3.2 [basic.def.odr] / 5 of ISO 14882:2003 for full details.
Yes. Usually, people put the declaration part in .h file and then put the function body implementations in .cpp file
Put the declarations in a .h file, then implement in as many files as you want, but include the .h in each.
But, you cannot implement the same thing in more than one file
I know this is a late post, but a friend just asked me a similar question, so I thought I would post our discussion:
From what I understand you want to do ThinkingStiff, it's not possible in C++. I take it you want something like "partial class" in C#.
Keep in mind, "partial class" in .NET is necessary so you don't have to put your entire implementation in one big source file. Unlike C++, .NET does not allow your implementation to be separate from the declaration, therefore the need for "partial class". So, Rafid it's not a nice feature, it's a necessary one, and a reinvention of the wheel IMHO.
If your class definition is so large you need to split it across more than one source file, then it can probably be split into smaller, decoupled classes and the primary class can use the mediator design pattern that ties them all together.
If your desire is to hide private members, such as when developing a commercial library and you don't wish to give away intelectual property (or hints to that affect), then you can use pure abstract classes. This way, all that is in the header is simple class with all the public definitions (API) and a factory function for creating an instance of it. All the implementation details are in your source file(s).
Another alternative is similar to the above, but uses a normal class (not a pure virtual one), but still exposes only the public interface. In your source file(s), you do all the work inside a nameless namespace, which can include friend functions from your class definition when necessary, etc. A bit more work, but a reasonable technique.
I bolded those things you might want to lookup and consider when solving your problem.
This is one of the nice features of C#, but unfortunately it is not supported in C++.
Perhaps something like policy based design would do the trick? Multiple Inheritience gives you the possiblity to compose classes in ways you can't in other languages. of course there are gotchas associated with MI so make sure you know what you are doing.
The short answer is yes, it is possible. The correct answer is no, you should never even think about doing it (and you will be flogged by other programmers should you disregard that directive).
You can split the definitions of the member functions into other translation units, but not of the class itself. To do that, you'd need to do preprocessor trickery.
// header file with function declarations
class C {
public:
void f(int foo, int bar);
int g();
};
// this goes into a seperate file
void C::f(int foo, int bar) {
// code ...
}
// this in yet another
void C::g() {
// code ...
}
There must be a reason why you want to split between two headers, so my guess is that you really want 2 classes that are joined together to form components of a single (3rd) class.
This question already has answers here:
Is the PIMPL idiom really used in practice?
(12 answers)
Closed 8 years ago.
Backgrounder:
The PIMPL Idiom (Pointer to IMPLementation) is a technique for implementation hiding in which a public class wraps a structure or class that cannot be seen outside the library the public class is part of.
This hides internal implementation details and data from the user of the library.
When implementing this idiom why would you place the public methods on the pimpl class and not the public class since the public classes method implementations would be compiled into the library and the user only has the header file?
To illustrate, this code puts the Purr() implementation on the impl class and wraps it as well.
Why not implement Purr directly on the public class?
// header file:
class Cat {
private:
class CatImpl; // Not defined here
CatImpl *cat_; // Handle
public:
Cat(); // Constructor
~Cat(); // Destructor
// Other operations...
Purr();
};
// CPP file:
#include "cat.h"
class Cat::CatImpl {
Purr();
... // The actual implementation can be anything
};
Cat::Cat() {
cat_ = new CatImpl;
}
Cat::~Cat() {
delete cat_;
}
Cat::Purr(){ cat_->Purr(); }
CatImpl::Purr(){
printf("purrrrrr");
}
I think most people refer to this as the Handle Body idiom. See James Coplien's book Advanced C++ Programming Styles and Idioms. It's also known as the Cheshire Cat because of Lewis Caroll's character that fades away until only the grin remains.
The example code should be distributed across two sets of source files. Then only Cat.h is the file that is shipped with the product.
CatImpl.h is included by Cat.cpp and CatImpl.cpp contains the implementation for CatImpl::Purr(). This won't be visible to the public using your product.
Basically the idea is to hide as much as possible of the implementation from prying eyes.
This is most useful where you have a commercial product that is shipped as a series of libraries that are accessed via an API that the customer's code is compiled against and linked to.
We did this with the rewrite of IONA's Orbix 3.3 product in 2000.
As mentioned by others, using his technique completely decouples the implementation from the interface of the object. Then you won't have to recompile everything that uses Cat if you just want to change the implementation of Purr().
This technique is used in a methodology called design by contract.
Because you want Purr() to be able to use private members of CatImpl. Cat::Purr() would not be allowed such an access without a friend declaration.
Because you then don't mix responsibilities: one class implements, one class forwards.
For what is worth, it separates the implementation from the interface. This is usually not very important in small size projects. But, in large projects and libraries, it can be used to reduce the build times significantly.
Consider that the implementation of Cat may include many headers, may involve template meta-programming which takes time to compile on its own. Why should a user, who just wants to use the Cat have to include all that? Hence, all the necessary files are hidden using the pimpl idiom (hence the forward declaration of CatImpl), and using the interface does not force the user to include them.
I'm developing a library for nonlinear optimization (read "lots of nasty math"), which is implemented in templates, so most of the code is in headers. It takes about five minutes to compile (on a decent multi-core CPU), and just parsing the headers in an otherwise empty .cpp takes about a minute. So anyone using the library has to wait a couple of minutes every time they compile their code, which makes the development quite tedious. However, by hiding the implementation and the headers, one just includes a simple interface file, which compiles instantly.
It does not necessarily have anything to do with protecting the implementation from being copied by other companies - which wouldn't probably happen anyway, unless the inner workings of your algorithm can be guessed from the definitions of the member variables (if so, it is probably not very complicated and not worth protecting in the first place).
If your class uses the PIMPL idiom, you can avoid changing the header file on the public class.
This allows you to add/remove methods to the PIMPL class, without modifying the external class's header file. You can also add/remove #includes to the PIMPL too.
When you change the external class's header file, you have to recompile everything that #includes it (and if any of those are header files, you have to recompile everything that #includes them, and so on).
Typically, the only reference to a PIMPL class in the header for the owner class (Cat in this case) would be a forward declaration, as you have done here, because that can greatly reduce the dependencies.
For example, if your PIMPL class has ComplicatedClass as a member (and not just a pointer or reference to it) then you would need to have ComplicatedClass fully defined before its use. In practice, this means including file "ComplicatedClass.h" (which will also indirectly include anything ComplicatedClass depends on). This can lead to a single header fill pulling in lots and lots of stuff, which is bad for managing your dependencies (and your compile times).
When you use the PIMPL idiom, you only need to #include the stuff used in the public interface of your owner type (which would be Cat here). Which makes things better for people using your library, and means you don't need to worry about people depending on some internal part of your library - either by mistake, or because they want to do something you don't allow, so they #define private public before including your files.
If it's a simple class, there's usually isn't any reason to use a PIMPL, but for times when the types are quite big, it can be a big help (especially in avoiding long build times).
Well, I wouldn't use it. I have a better alternative:
File foo.h
class Foo {
public:
virtual ~Foo() { }
virtual void someMethod() = 0;
// This "replaces" the constructor
static Foo *create();
}
File foo.cpp
namespace {
class FooImpl: virtual public Foo {
public:
void someMethod() {
//....
}
};
}
Foo *Foo::create() {
return new FooImpl;
}
Does this pattern have a name?
As someone who is also a Python and Java programmer, I like this a lot more than the PIMPL idiom.
Placing the call to the impl->Purr inside the .cpp file means that in the future you could do something completely different without having to change the header file.
Maybe next year they discover a helper method they could have called instead and so they can change the code to call that directly and not use impl->Purr at all. (Yes, they could achieve the same thing by updating the actual impl::Purr method as well, but in that case you are stuck with an extra function call that achieves nothing but calling the next function in turn.)
It also means the header only has definitions and does not have any implementation which makes for a cleaner separation, which is the whole point of the idiom.
We use the PIMPL idiom in order to emulate aspect-oriented programming where pre, post and error aspects are called before and after the execution of a member function.
struct Omg{
void purr(){ cout<< "purr\n"; }
};
struct Lol{
Omg* omg;
/*...*/
void purr(){ try{ pre(); omg-> purr(); post(); }catch(...){ error(); } }
};
We also use a pointer-to-base class to share different aspects between many classes.
The drawback of this approach is that the library user has to take into account all the aspects that are going to be executed, but only sees his/her class. It requires browsing the documentation for any side effects.
I just implemented my first PIMPL class over the last couple of days. I used it to eliminate problems I was having, including file *winsock2.*h in Borland Builder. It seemed to be screwing up struct alignment and since I had socket things in the class private data, those problems were spreading to any .cpp file that included the header.
By using PIMPL, winsock2.h was included in only one .cpp file where I could put a lid on the problem and not worry that it would come back to bite me.
To answer the original question, the advantage I found in forwarding the calls to the PIMPL class was that the PIMPL class is the same as what your original class would have been before you pimpl'd it, plus your implementations aren't spread over two classes in some weird fashion. It's much clearer to implement the public members to simply forward to the PIMPL class.
Like Mr Nodet said, one class, one responsibility.
I don't know if this is a difference worth mentioning but...
Would it be possible to have the implementation in its own namespace and have a public wrapper / library namespace for the code the user sees:
catlib::Cat::Purr(){ cat_->Purr(); }
cat::Cat::Purr(){
printf("purrrrrr");
}
This way all library code can make use of the cat namespace and as the need to expose a class to the user arises a wrapper could be created in the catlib namespace.
I find it telling that, in spite of how well-known the PIMPL idiom is, I don't see it crop up very often in real life (e.g., in open source projects).
I often wonder if the "benefits" are overblown; yes, you can make some of your implementation details even more hidden, and yes, you can change your implementation without changing the header, but it's not obvious that these are big advantages in reality.
That is to say, it's not clear that there's any need for your implementation to be that well hidden, and perhaps it's quite rare that people really do change only the implementation; as soon as you need to add new methods, say, you need to change the header anyway.