Related
Recently I have fallen in a situation like this. I'm generalizing the problem because I think it relates more to the structural design than the specific problem.
General problem
There is a hierarchy of classes: an abstract base class Base and some concretions D1, D2, D3 that inherit from it. The class A contains an object's collection of type Base. A requires a computation from some service-class B but B.process() method accepts only a collection of type D1. Let's say that is important because if the input collection contains any other type the value returned is just wrong.
A have an interface that allows clients to add elements to the internal collection, which is not exposed in any other way. The classes in the hierarchy can be constructed for the same clients and pass the new values to A; A have not enough context to construct them itself.
Attempts, questions and thoughts
The major concern for me was the need to determine at runtime the type of each element in the A collection, so can filter the right ones and pass to B.process(). Even if it is possible (it is in my particular problem, more later on) it just seems wrong! I think the object who contains references to the abstract base class shouldn't have to know the concrete instances it holds.
I try to:
Change the parameter type to B.process(c: Base[]) so A doesn't have to downcast the type, but it doesn't solve anything: A still needs to filter the elements or the computation will be wrong.
Pass the complete collection Base[] to B.process() but just defer the problem of selection/downcasting to B.
Put a process() method in Base so D1 can override the behavior (well known polymorphism). The problem here is that a process() returning a SomeValue type just have sense for D1.
Separate the interface that add elements so a more specific A.addD1Element(e: D1) method could allow put D1 objects in a different collection and pass that to B. It should work but also looks... don't know, weird. If method overload based on parameter type is possible at least the process won't be so cumbersome for clients of the class.
Just separate the D1 class of the hierarchy. This is a more aggressive variation of the previous one. The issue is that D1 seems related to the whole hierarchy except for the specific requirements of B.
Those were some of my thoughts on the problem.
For instance, the language used have support to check the type of an object at runtime (instanceof) and it is easy to filter the collection based on that check. But as I say my question is more related to the paradigm. What about a language, say for instance C++, where is less handy to make a check like that?
So what could be a solution to this kind of problem? What kind of refactoring or design pattern could be applied so the problem is easy to treat with or simply fades away?
This question looks related, but I believe this is more general (although I provide a more specific context). The most upvoted answer suggest to split in different collections. This is also a think i'm considering, but that forces to change A implementation every time a new type is added.
Context (problem in action)
I'm asking in a general way because it really intrigues me on that way, but I know most of the time a design can be analyzed only with the context of the particular problem it tries to solve.
The problem at hand is similar to this:
A is a class (some kind of entity, like a DDD entity) that models a sort of agreement or debt a customer incurs for a service. It has different costs including a monthly pay. Base and related classes are Payments of different types. They share a lot in common, although most of it is data (date, amount, interests, etc); but there is at least one type of payment that have different, additional information: the monthly payment (D1). Those payments need to be analyzed carefully so a different class (B) is responsible for that, using more contextual information and all the payments of that type at once. The service needs the additional data that is specific to those payments so cannot receive an abstract Payment type (at least not in that design). Other payments doesn't have the specific information MonthlyPayment does and so they cannot generates the values that business requires and B is generating (doesn't have sense in other payment types).
All payments are stored in the same collection so other methods of the class can process all payments in a generic way.
This is mostly the context. I think the design is not the best, but I fail to see a better one.
Maybe separating only MonthlyPayment (D1) in a different collection as described earlier? But it is not the only payment that requires additional processing (it is the most complex, though), so I could end with different collections for every payment type and no hierarchy at all. Right now there are four payments types and two of them requires additional, specific analysis, but more types can be added later and the issue of need to modify the implementation every time a new type is added persists.
Is this, more discrete approach of different collections by type, a better one here? The abstract base class Payment can still be used for payments that can be manipulated trough the common interface. Also I can use a layer super type or something like that to allow reutilization of common functionality (the language allows a kind of mixing as well) and stop using the base class as root from a hierarchy.
Uf. I am sorry for the length of the text. I hope it is at least readable and clear. Thank you very much in advance.
I'll give you an example about path finding. When you wnat to find a path, you can pick a final destination, a initial position and find the fastest way between the two, or you can just define the first position, and let the algorithm show every path you can finish, or you may want to mock this for a test and just say the final destination and assume you "teleport" to there, and so on. It's clear that the function is the same: finding a path. But the arguments may vary between implementations. I've searched a lot and found a lot of solutions: getting rid of the interface, putting all the arguments as fields in the implementation, using the visitor pattern...
But I'd like to know from you guys what is the drawback of putting every possible argument (not state) in one object (let's call it MovePreferences) and letting every implementation take what it needs. Sure, may you need another implementation that takes as argument that you didn't expect, you will need to change the MovePreferences, but it don't sound too bad, since you will only add methods to it, not refactor any existing method. Even though this MovePreferences is not an object of my domain, I'm still tempted to do it. What do you think?
(If you have a better solution to this problem, feel free to add it to your answer.)
The question you are asking is really why have interfaces at all, no, why have any concept of context short of 'whatever I need?' I think the answers to that are pretty straightforward: programming with shared global state is easy for you, the programmer, and quickly turns into a vortex for everyone else once they have to coalesce different features, for different customers, render enhancements, etc.
Now the far other end of the spectrum is the DbC argument: every single interface must be a highly constrained contract that not only keeps the knowledge exchanged to an absolute minimum, but makes the possibility of mayhem minimal.
Frankly, this is one of the reasons why dependency injection can quickly turn into a mess: as soon as design issues like this come up, people just start injecting more 'objects,' often to get access to just one property, whose scope might not be the same as the scope of the present operation. [Different kind of nightmare.]
Unfortunately, there's almost no information in your question. Do I think it would be possible to correctly model the notion of a Route? Sure. That doesn't sound very challenging. Here are a few ideas:
Make a class called Route that has starting and ending points. Then a collection of Traversals. The idea here would be that a Route could completely ignore the notion of how someone got from point a to point b, where traversal could contain information about roads, traffic, closures, whatever. Then your mocked case could just have no Traversals inside.
Another option would be to make Route a Composite so that each trip is then seen as the stringing together of various segments. That's the way routes are usually presented: go 2 miles on 2 South, exit, go 3 miles east on Santa Monica Boulevard, etc. In this scenario, you could just have Routes that have no children.
Finally, you will probably need a creational pattern. Perhaps a Builder. That simplifies mocking things too because you can just make a mock builder and have it construct Routes that consist of whatever you need.
The other advantage of combining the Composite and Builder is that you could make a builder that can build a new Route from an existing one by trying to improve only the troubling subsegments, e.g. it got traffic information that the 2S was slow, it could just replace that one segment and present its new route.
Consider an example,
Say if 5 arguments are encapsulated in an object and passed on to 3 methods.
If the object undergoes change in structure, then we need to run test cases for all the 3 methods. Instead if the method accepts only the arguments they need, they need not be tested.
Only problem I see out of this is Increase in Testing Efforts
Secondly you will naturally violate Single Responsibility Principle(SRP) if you pass more arguments than what the method actually needs.
Reading the wikipedia entry about God Objects, it says that a class is a god object when it knows too much or does too much.
I see the logic behind this, but if it's true, then how do you couple every different class? Don't you always use a master class for connecting window management, DB connections, etc?
The main function/method may know about the existence of the windows, databases, and other objects. It may perform over-arching tasks like introduce the model to the controller.
But that doesn't mean it manages all the little details. It probably doesn't know anything about how the database or windows are implemented.
If it did, it could be accused of being a God object.
A god object is an object that contains references, directly or indirectly, to most if not all objects within an application. As the question observes, it is almost impossible to avoid having a god object in an application. Some object must hold references to the various subsystems: UI, database, communications, business logic, etc. Note that the god object need not be application-defined. Many frameworks have built-in god objects with names like "application context", "application environment", "session", "activator", etc.
The issue is not whether a god object exists, but rather how it is used. I will illustrate with an extreme example...
Let's say that in my application I want to standardize how many decimal places of precision to show when displaying numbers. However, I want the precision to be configurable. I create a class whose responsibility is to convert numbers to strings:
class NumberFormatter {
...
String format(double value) {
int decimalPlaces = getConfiguredPrecision();
return formatDouble(value, decimalPlaces);
}
int getConfiguredPrecision() {
return /* what ??? */;
}
}
The question is, how does getConfiguredPrecision figure out what to return? One way would be to give NumberFormatter a reference to the global application context which it stores in a member field called _appContext. Then we could write:
return _appContext.getPreferenceManager().getNumericPreferences().getDecimalPlaces();
By doing this, we have just made NumberFormatter into a god object as well! Why? Because now we can (indirectly) reference virtually any object in the application through its _appContext field. Is this bad? Yes, it is.
I'm going to write a unit test for NumberFormatter. Let's set up the parameters... it needs an application context?! WTF, that has 57 methods I need to mock. Oh, it only needs the pref manager... WTF, I have to mock 14 methods! Numeric prefs!?! Screw it, the class is simple enough, I don't need to test it...
Let's say that the application context had another method, getDatabaseManager(). Last week we were using SQL, so the method returned an SQL database object. But this week, we've decided to change to a NoSQL database and the method now returns a new type. Is NumberFormatter affected by the change? Hmmm, I can't remember... yeah, it might be, I see it takes an application context in the constructor... let me open the source and take a look... nope, we're in luck: it only accesses getPreferenceManager()... now let's check the other 93 classes that take an application context as a parameter...
This same scenario occurs if a change is made to the preferences manager, or the numeric preferences object. The moral of the story is that an object should only hold references to the things that it needs to perform its job, and only those things. In the case of NumberFormatter, all it needs to know is a single integer -- the number of decimal places. It could be created directly by the application god object who knows the magic number (or the pref manager or better still, numeric prefs), without turning the formatter into a god object itself. Furthermore, any components that need to format numbers could be given a formatter instead of the god object. Wins all around.
So, to summarize, the problem is not the existence of a god object but rather the act of conferring god-like status to other objects willy-nilly.
Incidentally, the design principle that tackles this problem head-on has become known as the Law of Demeter. Or "when paying at a restaurant, give the server your money not your wallet."
In my experience this most often occurs when you're dealing with code that is the product of "Develop as you go" project management (or lack there of). When a project is not thought through and planned and object responsibilities are loose and not delegated properly. In theses scenarios you find a "god-object" being the catchall for code that doesn't have any obvious organization or delegation.
It is not the interconnectedness or coupling of the different classes that is the problem with god-objects, it's the fact that a god-object many times can accomplish most if not all responsibilities of it's derived children, and are fairly unpredictable (by anyone other than the developer) as to what their defined responsibilities are.
Simply knowing about "multiple" classes doesn't make one a God; knowing about multiple classes in order to solve a problem that should be split into several sub-problems does make one a God.
I think the focus should be on whether a problem should be split into several sub-problems, not on the number of classes a given object knows about (as you pointed out, sometimes knowing about several classes is necessary).
Gods are over-hyped.
I often see two conflicting strategies for method interfaces, loosely summarized as follows:
// Form 1: Pass in an object.
double calculateTaxesOwed(TaxForm f) { ... }
// Form 2: Pass in the fields you'll use.
double calculateTaxesOwed(double taxRate, double income) { ... }
// use of form 1:
TaxForm f = ...
double payment = calculateTaxesOwed(f);
// use of form 2:
TaxForm f = ...
double payment = calculateTaxesOwed(f.getTaxRate(), f.getIncome());
I've seen advocates for the second form, particularly in dynamic languages where it may be harder to evaluate what fields are being used.
However, I much prefer the first form: it's shorter, there is less room for error, and if the definition of the object changes later you won't necessarily need to update method signatures, perhaps just change how you work with the object inside the method.
Is there a compelling general case for either form? Are there clear examples of when you should use the second form over the first? Are there SOLID or other OOP principles I can point to to justify my decision to use one form over the other? Do any of the above answers change if you're using a dynamic language?
In all honesty it depends on the method in question.
If the method makes sense without the object, then the second form is easier to re-use and removes a coupling between the two classes.
If the method relies on the object then fair enough pass the object.
There is probably a good argument for a third form where you pass an interface designed to work with that method. Gives you the clarity of the first form with the flexibility of the second.
It depends on the intention of your method.
If the method is designed to work specifically with that object and only that object, pass the object. It makes for a nice encapsulation.
But, if the method is more general purpose, you will probably want to pass the parameters individually. That way, the method is more likely to be reused when the information is coming from another source (i.e. different types of objects or other derived data).
I strongly recommend the second solution - calculateTaxesOwed() calculates some data, hence needs some numerical input. The method has absolutly nothing to do with the user interface and should in turn not consum a form as input, because you want your business logic separated from your user interface.
The method performing the calculation should (usualy) not even belong to the same modul as the user interface. In this case you get a circular dependency because the user interface requires the business logic and the business logic requires the user interface form - a very strong indication that something is wrong (but could be still solved using interface based programming).
UPDATE
If the tax form is not a user interface form, things change a bit. In this case I suggest to expose the value using a instance method GetOwedTaxes() or instance property OwedTaxes of the TaxForm class but I would not use a static method. If the calculation can be reused elsewhere, one could still create a static helper method consuming the values, not the form, and call this helper method from within the instance method or property.
I don't think it really matters. You open yourself to side effects if you pass in the Object as it might be mutated. This might however be what you want. To mitigate this (and to aid testing) you are probably better passing the interface rather than the concrete type. The benefit is that you don't need to change the method signature if you want to access another field of the Object.
Passing all the parameters makes it clearer what the type needs, and might make it easier to test (though if you use the interface this is less of a benefit). But you will have more refactoring.
Judge each situation on its merits and pick the least painful.
Passing just the arguments can be easier to unit test, as you don't need to mock up entire objects full of data just to test functionality that is essentially just static calculation. If there are just two fields being used, of the object's many, I'd lean towards just passing those fields, all else being equal.
That said, when you end up with six, seven or more fields, it's time to consider passing either the whole object or a subset of the fields in a "payload" class (or struct/dictionary, depending on the language's style). Long method signatures are usually confusing.
The other option is to make it a class method, so you don't have to pass anything. It's less convenient to test, but worth considering when your method is only ever used on a TaxForm object's data.
I realize that this is largely an artifact of the example used and so it may not apply in many real-world cases, but, if the function is tied so strongly to a specific class, then shouldn't it be:
double payment = f.calculateTaxesOwed;
It seems more appropriate to me that a tax document would carry the responsibility itself for calculating the relevant taxes rather than having that responsibility fall onto a utility function, particularly given that different tax forms tend to use different tax tables or calculation methods.
One advantage of the first form is
Abstraction - programming to an interface rather than implementation. It makes the maintainance of your code easier in the long run becuase you may change the implementation of TaxForm without affecting the client code as long as the interface of TaxForm does not change.
This is the same as the "Introduce Parameter Object" from Martin Fowler's book on refactoring. Fowler suggests that you perform this refactoring if there are a group of parameters that tend to be passed together.
If you believe in the Law of Demeter, then you would favor passing exactly what is needed:
http://en.wikipedia.org/wiki/Law_of_Demeter
http://www.c2.com/cgi/wiki?LawOfDemeter
Separation of UI and Data to be manipulated
In your case, you are missing an intermediate class, say, TaxInfo, representing the entity to be taxed. The reason is that UI (the form) and business logic (how tax rate is calculated) are on two different "change tracks", one changes with presentation technology ("the web", "The web 2.0", "WPF", ...), the other changes with legalese. Define a clear interface between them.
General discussion, using an example:
Consider a function to create a bitmap for a business card. Is the purpose of the function
(1) // Formats a business card title from first name and last name
OR
(2) // Formats a businnes card title from a Person record
The first option is more generic, with a weaker coupling, which is generally preferrable. However, In many cases less robust against change requests - e.g. consider "case 2017: add persons Initial to business card".
Changing the implementation (adding person.Initial) is usually easier and faster than changing the interface.
The choice is ultimately what type of changes you expect: is it more likely that more information from a Personrecord is required, or is it more likely that you want to create business card titles for other data structures than Person?
If that is "undecided", anfd you can't opf for purpose (1) or (2) I'd rather go with (2), for syntactic cleanliness.
If I was made to choose one of the two, I'd always go with the second one - what if you find that you (for whatever reason) need to caculate the taxes owed, but you dont have an instance of TaxForm?
This is a fairly trivial example, however I've seen cases where a method doing a relatively simple task had complex inputs which were difficult to create, making the method far more difficult to use than it should have been. (The author simply hadn't considered that other people might want to use that method!)
Personally, to make the code more readable, I would probbaly have both:
double calculateTaxesOwed(TaxForm f)
{
return calculateTaxesOwed(f.getTaxRate(), f.getIncome());
}
double calculateTaxesOwed(double taxRate, double income) { ... }
My rule of thumb is to wherever possible have a method that takes exactly the input it needs - its very easy to write wrapper methods.
Personally, I'll go with #2 since it's much more clear of what it is that the method need. Passing the TaxForm (if it is what I think it is, like a Windows Form) is sort of smelly and make me cringe a little (>_<).
I'd use the first variation only if you are passing a DTO specific to the calculation, like IncomeTaxCalculationInfo object which will contain the TaxRate and Income and whatever else needed to calculate the final result in the method, but never something like a Windows / Web Form.
I often find myself needing reference to an object that is several objects away, or so it seems. The options I see are passing a reference through a middle-man or just making something available statically. I understand the danger of global scope, but passing a reference through an object that does nothing with it feels ridiculous. I'm okay with a little bit passing around, I suppose. I suspect there's a line to be drawn somewhere.
Does anyone have insight on where to draw this line?
Or a good way to deal with the problem of distributing references amongst dependent objects?
Use the Law of Demeter (with moderation and good taste, not dogmatically). If you're coding a.b.c.d.e, something IS wrong -- you've nailed forevermore the implementation of a to have a b which has a c which... EEP!-) One or at the most two dots is the maximum you should be using. But the alternative is NOT to plump things into globals (and ensure thread-unsafe, buggy, hard-to-maintain code!), it is to have each object "surface" those characteristics it is designed to maintain as part of its interface to clients going forward, instead of just letting poor clients go through such undending chains of nested refs!
This smells of an abstraction that may need some improvement. You seem to be violating the Law of Demeter.
In some cases a global isn't too bad.
Consider, you're probably programming against an operating system's API. That's full of globals, you can probably access a file or the registry, write to the console. Look up a window handle. You can do loads of stuff to access state that is global across the whole computer, or even across the internet... and you don't have to pass a single reference to your class to access it. All this stuff is global if you access the OS's API.
So, when you consider the number of global things that often exist, a global in your own program probably isn't as bad as many people try and make out and scream about.
However, if you want to have very nice OO code that is all unit testable, I suppose you should be writing wrapper classes around any access to globals whether they come from the OS, or are declared yourself to encapsulate them. This means you class that uses this global state can get references to the wrappers, and they could be replaced with fakes.
Hmm, anyway. I'm not quite sure what advice I'm trying to give here, other than say, structuring code is all a balance! And, how to do it for your particular problem depends on your preferences, preferences of people who will use the code, how you're feeling on the day on the academic to pragmatic scale, how big the code base is, how safety critical the system is and how far off the deadline for completion is.
I believe your question is revealing something about your classes. Maybe the responsibilities could be improved ? Maybe moving some code would solve problems ?
Tell, don't ask.
That's how it was explained to me. There is a natural tendency to call classes to obtain some data. Taken too far, asking too much, typically leads to heavy "getter sequences". But there is another way. I must admit it is not easy to find, but improves gradually in a specific code and in the coder's habits.
Class A wants to perform a calculation, and asks B's data. Sometimes, it is appropriate that A tells B to do the job, possibly passing some parameters. This could replace B's "getName()", used by A to check the validity of the name, by an "isValid()" method on B.
"Asking" has been replaced by "telling" (calling a method that executes the computation).
For me, this is the question I ask myself when I find too many getter calls. Gradually, the methods encounter their place in the correct object, and everything gets a bit simpler, I have less getters and less call to them. I have less code, and it provides more semantic, a better alignment with the functional requirement.
Move the data around
There are other cases where I move some data. For example, if a field moves two objects up, the length of the "getter chain" is reduced by two.
I believe nobody can find the correct model at first.
I first think about it (using hand-written diagrams is quick and a big help), then code it, then think again facing the real thing... Then I code the rest, and any smells I feel in the code, I think again...
Split and merge objects
If a method on A needs data from C, with B as a middle man, I can try if A and C would have some in common. Possibly, A or a part of A could become C (possible splitting of A, merging of A and C) ...
However, there are cases where I keep the getters of course.
But it's less likely a long chain will be created.
A long chain will probably get broken by one of the techniques above.
I have three patterns for this:
Pass the necessary reference to the object's constructor -- the reference can then be stored as a data member of the object, and doesn't need to be passed again; this implies that the object's factory has the necessary reference. For example, when I'm creating a DOM, I pass the element name to the DOM node when I construct the DOM node.
Let things remember their parent, and get references to properties via their parent; this implies that the parent or ancestor has the necessary property. For example, when I'm creating a DOM, there are various things which are stored as properties of the top-level DomDocument ancestor, and its child nodes can access those properties via the reference which each one has to its parent.
Put all the different things which are passed around as references into a single class, and then pass around just that one class instance as the only thing that's passed around. For example, there are many properties required to render a DOM (e.g. the GDI graphics handle, the viewport coordinates, callback events, etc.) ... I put all of these things into a single 'Context' instance which is passed as the only parameter to the methods of the DOM nodes to be rendered, and each method can get whichever properties it needs out of that context parameter.