With extension methods we can easily add methods to any type. Obviously this opens the possibility in a future version of .net the extension method could no longer get called (for example the type now includes a method with identical signature to the extension method).
Should this be a concern?
If so, how should I deal with this and design my extension methods as to minimise code changes should this happen?
If the framework is changed so much in the future, there will always be compatibility issues. If a new framework method is added with the same name as your extension method, it is quite likely that they have the same, or at least very similar functionality and a refactoring is due anyways.
I think that the power of the extension methods is too large to ignore just because of this risk.
I'm afraid that the only thing you can do is providing unique enough names to your extension methods so you're 100% sure you will never have a conflict.
Not talking about adding the name of your cat to the method's name, just try to be more creative :)
Use obscure method names that would never be used in the framework.
edit -- perhaps obscure wasn't the most appropriate word, please substitute with meaningful but less common verbage
Attempting to avoid signature conflicts is really the only strategy to avoiding the hassle of code rework (assuming the functionality of the extension method needs to be preserved and not simply converted to the framework's definition of the method).
Related
As shown here, attribute constructors are not called until you reflect to get the attribute values. However, as you may also know, you can only pass compile-time constant values to attribute constructors. Why is this? I think many people would much prefer to do something like this:
[MyAttribute(new MyClass(foo, bar, baz, jQuery)]
than passing a string (causing stringly typed code too!) with those values, turned into strings, and then relying on Regex to try and get the value instead of just using the actual value, and instead of using compile-time warnings/errors depending on exceptions that might be thrown somewhere that has nothing to do with the class except that a method that it called uses some attributes that were typed wrong.
What limitation caused this?
Attributes are part of metadata. You need to be able to reflect on metadata in an assembly without running code in that assembly.
Imagine for example that you are writing a compiler that needs to read attributes from an assembly in order to compile some source code. Do you really want the code in the referenced assembly to be loaded and executed? Do you want to put a requirement on compiler writers that they write compilers that can run arbitrary code in referenced assemblies during the compilation? Code that might crash, or go into infinite loops, or contact databases that the developer doesn't have permission to talk to? The number of awful scenarios is huge and we eliminate all of them by requiring that attributes be dead simple.
The issue is with the constructor arguments. They need to come from somewhere, they are not supplied by code that consumes the attribute. They must be supplied by the Reflection plumbing when it creates the attribute object by calling its constructor. For which it needs the constructor argument values.
This starts at compile time with the compiler parsing the attribute and recording the constructor arguments. It stores those argument values in the assembly metadata in a binary format. At issue then is that the runtime needs a highly standardized way to deserialize those values, one that preferably doesn't depend on any of the .NET classes that you'd normally use the de/serialize data. Because there's no guarantee that such classes are actually available at runtime, they won't be in a very trimmed version of .NET like the Micro Framework.
Even something as common as binary serialization with the BinaryFormatter class is troublesome, note how it requires the [Serializable] attribute on the class to allow it to do its job. Versioning would also be an enormous problem, clearly such a serializer class could never change for the risk of breaking attributes in old assemblies.
This is a rock and a hard place, solved by the CLS designers by heavily restricting the allowed types for an attribute constructor. They didn't leave much, just the simple values types, string, a simple one-dimensional array of them and Type. Never a problem deserializing them since their binary representation is simple and unambiguous. Quite a restriction but attributes can still be pretty expressive. The ultimate fallback is to use a string and decode that string in the constructor at runtime. Creating an object of MyClass isn't an issue, you can do so in the attribute constructor. You'll have to encode the arguments that this constructor needs however as properties of the attribute.
The probably most correct answer as to why you can only use constants for attributes is because the C#/BCL design team did not judge supporting anything else important enough to be added (i.e. not worth the effort).
When you build, the C# compiler will instantiate the attributes you have placed in your code and serialize them, so that they can be stored in the generated assembly. It was probably more important to ensure that attributes can be retrieved quickly and reliably than it was to support more complex scenarios.
Also, code that fails because some attribute property value is wrong is much easier to debug than some framework-internal deserialization error. Consider what would happen if the class definition for MyClass was defined in an external assembly - you compile and embed one version, then update the class definition for MyClass and run your application: boom!
On the other hand, it's seriously frustrating that DateTime instances are not constants.
What limitation caused this?
The reason it isn't possible to do what you describe is probably not caused by any limitation, but it's purely a language design decision. Basically, when designing the language they said "this should be possible but not this". If they really wanted this to be possible, the "limitations" would have been dealt with and this would be possible. I don't know the specific reasoning behind this decision though.
/.../ passing a string (causing stringly typed code too!) with those values, turned into strings, and then relying on Regex to try and get the value instead of just using the actual value /.../
I have been in similar situations. I sometimes wanted to use attributes with lambda expressions to implement something in a functional way. But after all, c# is not a functional language, and if I wrote the code in a non-functional way I haven't had the need for such attributes.
In short, I think like this: If I want to develop this in a functional way, I should use a functional language like f#. Now I use c# and I do it in a non-functional way, and then I don't need such attributes.
Perhaps you should simply reconsider your design and not use the attributes like you currently do.
UPDATE 1:
I claimed c# is not a functional language, but that is a subjective view and there is no rigourous definition of "Functional Language". I agree with the Adam Wright, "/.../ As such, I wouldn't class C# as functional in general discussion - it is, at best, multi-paradigm, with some functional flavour." at Why is C# a functional programmming language?
UPDATE 2:
I found this post by Jon Skeet: https://stackoverflow.com/a/294259/1105687 It regards not allowing generic attribute types, but the reasoning could be similar in this case:
Answer from Eric Lippert (paraphrased): no particular reason, except
to avoid complexity in both the language and compiler for a use case
which doesn't add much value.
I have a Web solution which contains two projects (A and B) with B referencing A.
In A I have an Html extension method that obviously can be called from either A or B.
My question is once the method is called (usually from a partial view) is there a way inside the method to figure out whether the call came from Assembly A or Assembly B without passing anything to it?
I tried to see if I can do anything with HttpContext.Current.Request but could not find anything useful. I can get the URI but that still does not tell me which assembly the file that originated the Request is in.
Thanks for your answers - the method returns a string and the string is from a string.resx file which I have one for each assembly. That is why I need to know which file to access to return the string. Since each assembly "registers" itself on start up if I add a new assembly my method will not change, since it will just look up the assembly.In fact my whole project will not change. The reason why I am not introducing another parameter at this time is b/c it will mean a HUGE amount of changes and I honestly don't see the benefit. While I see your point and I generally agree with it I think in my case it's not that the method returns different things , it's just grabbing the correct resource file based on the assembly.
As SLaks pointed out, you can check HttpContext.Current.Application.GetType().Assembly.
However I agree with John in the comments that you have probably made a bad design decision if you need this.
The Problem
Your method is a hypocrite.
It talks different to different callers but doesn't tell it in open.
You see, each method defines a certain contract with arguments and a return type.
For example, int.Parse says that it takes a string and turns it into an int. If we want to change default behavior, we may also give it NumberStyles and/or IFormatProvider.
We the consumers don't know how int.Parse is implemented. Because it is static, we most certainly expect it doesn't have side effects and will always return the same value for the same set of parameters.
Repeat this mantra after me:
Explicit is better than implicit.
You would probably be very angry if you found out int.Parse somehow analyzes your code and changes its behavior depending on where it's called from.
It's the caller's responsibility to define the context, not the callee's.
Try to give simple and concise answers to questions below:
What happens if the method is called from assembly C?
How would you unit-test it? What if some other developer uses this method in unit tests?
What happens if you rename assembly A or B? Merge them? Split them further?
Will you remember to change this method if anything above happens?
If answering any of the questions above clearly poses a challenge for you, it is a sign you're Doing It Wrong™.
Instead you should...
Introduce a Parameter
Think about the method contract. What can you do to make it full and descriptive?
Define a generic (as in English) method in a separate assembly that doesn't know anything about the callers and has additional parameters, and define parameter-filling shortcuts for it in concrete assemblies.
It's better that these parameters don't know anything about the assemblies either.
For example, if you needed to resolve URLs inside your method, you could accept string baseUrl or Func<string, string> urlResolver so it's potentially usable from any assembly that cares to specify those.
In the worst case, you could define an enum with possible caller contexts and pass it to the method. This will make your design problem explicit, rather than implicit. Obvious problem is always better than hidden problem, although worse than no problem at all.
Check HttpContext.Current.Application.GetType().Assembly
From what I have read about extension methods, you will run into problems if the base class decides to add a method with the same name as your extension. For a specific class it is generally not to hard to pick a name to avoid clashes, however extension methods can be added to interfaces which adds infinitely more potential for conflicts.
In Objective-C (with their version, categories), this problem is avoided by adding a prefix before each method. I know we can define extension methods in namespaces so that we can control whether they are imported or not, but this only solves the problem of clashes between extension methods, rather than clashes between an extension method and the base class.
Update:
Nobody, actually mentioned this, but extension methods aren't virtual. That means if you can i.myExtension() on an interface i, then it will always call the interface classes method, so the subclass method (which could have different intent) won't be called. So, overall, using extension methods is quite safe.
The use of a prefix will make the code ugly, which removes some of the value of an extension method.
I recommend instead, being careful to not create a large number of extension methods and to use method names that make sense. This way, in the case of a conflict, it's more likely that, for instance, the extension Where method and the conflicting Where method will have similar semantics.
The common convention is to have them in their own separate namespace (usually in the form of <something>.Extensions) which you can then use to decide whether or not you use them in any particular piece of code which contains types the extensions normally operate on.
Furthermore, if you do have a conflict, remember that extension methods are still static methods, so you can always call them explicitly in the case of a name conflict.
However, if you find that you are frequently running into name conflicts, you might want to reconsider the names you are choosing so that they don't interfere with names that already exist on the type.
To summarize, no, you should not prefix extension method names in C#, as they only serve to obfuscate your code and make it harder to maintain.
If the base class implements a method with the same name as your extension, I'd guess you'd have one of these two scenarios:
The base class implemented the same functionality you were adding yourself with an extension method. You no longer need the extension method. Delete it and use the method on the base class instead.
The base class implemented something different from what you wanted but gave it the same name. Either you or they are using the wrong name. Probably you. Rename your extension method to something that describes what the function actually does.
I'm inclined to say no they shouldn't unless you have reason good to suspect that the method signature you are using is going to be implemented. The reason i think this is...
It's is quite unlikely that your method signature will be duplicated in most cases.
If it does, it should always be detected in unit tests.
Should be trivial to fix.
I normally opt for relatively verbose method names when implementing extension methods anyway to avoid confusion.
Long explanation aside, I have a situation where I need to basically re-implement a .NET framework class in order to extend the behavior in a manner that is not compatible with an inheritance or composition/delegation strategy. The question is not a matter of whether the course of action I am to take is what you would do, or recommend, it is instead a question of naming/coding-style.
Is there a paradigm for naming classes and methods that have the same functionality as an existing class or method ala the convention of ClassEx/MethodEx that exists in C++?
[edit]
I understand that choosing good names for this is important... I haven't written a line of code yet, and am instead taking the time to think through the ramifications of what I am about to undertake, and that includes searching for a clear, descriptive, name while trying to be concise. The issue is that the name I have in mind is not terribly concise.
[/edit]
Here are the ways I've seen in the .NET Framework itself:
Call it something slightly different, but don't use any specific suffix. For example, System.TimeZoneInfo was introduced to supersede System.TimeZone.
Put it in another namespace. For example, the WPF Button is in System.Windows instead of System.Windows.Forms.
Suffix it with a number. For example X509Certificate2 versus X509Certificate. (This practice was common with COM interfaces but has fallen out of favor in .NET.)
Note that the naming of TimeZoneInfo is a publicized case of Microsoft tackling this convtrovertial naming issue head on. See and http://blogs.msdn.com/kathykam/archive/2007/03/28/bye-bye-system-timezone2-hello-system-timezoneinfo.aspx and http://blogs.msdn.com/kcwalina/archive/2006/10/06/TimeZone2Naming.aspx for excellent information.
Try name your classes/methods with real meaning.
For example, if you extending the Random functionality to create random strings, name the class StringRandom or StringRandomizer and such.
If you writing class with general purpose extension methods that applying to specific class/interface, for example IList, name it ListExtensions.
If you writing random.Next method that returns random number between minValue and maxValue including maxValue, name the method NextIncludingMaxValue.
If you writing queue.Dequeue method that is thread safe, name if DequeueThreadSafe.
If you writing queue.Dequeue method that blocking until other thread enqueueing an item, name it DequeueBlocking.
And such...
C#, for the most part, avoids these situations entirely due to the ease in which you can extend a class with new methods without breaking binary compatibility (you can add methods, at will, to a class, just not an interface), and through the use of Extension methods.
There are few reasons to ever do this in C#, unlike C++. In C++, adding a method breaks compatibility, so "Ex" becomes a much more common scenario.
I give all my methods (and properties) camelCase names: so for example Invalidate is a framework method name, and invalidate is the name of one of my methods.
This (using camelCase names) is unconventional, so some people object to it, but I find it convenient.
No such problem with class names (for which I use the conventional UpperCase), because for class names there are their namespaces to distinguish them from the framework classes.
Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 6 years ago.
Improve this question
In what circumstances (usage scenarios) would you choose to write an extension rather than sub-classing an object ?
< full disclosure : I am not an MS employee; I do not know Mitsu Furota personally; I do know the author of the open-source Componax library mentioned here, but I have no business dealings with him whatsoever; I am not creating, or planning to create any commercial product using extensions : in sum : this post is from pure intellectal curiousity related to my trying to (continually) become aware of "best practices" >
I find the idea of extension methods "cool," and obviously you can do "far-out" things with them as in the many examples you can in Mitsu Furota's (MS) blog postslink text.
A personal friend wrote the open-source Componax librarylink text, and there's some remarkable facilities in there; but he is in complete command of his small company with total control over code guidelines, and every line of code "passes through his hands."
While this is speculation on my part : I think/guess other issues might come into play in a medium-to-large software team situation re use of Extensions.
Looking at MS's guidelines at link text, you find :
In general, you will probably be
calling extension methods far more
often than implementing your own. ...
In general, we recommend that you
implement extension methods sparingly
and only when you have to. Whenever
possible, client code that must extend
an existing type should do so by
creating a new type derived from the
existing type. For more information,
see Inheritance (C# Programming
Guide). ... When the compiler
encounters a method invocation, it
first looks for a match in the type's
instance methods. If no match is
found, it will search for any
extension methods that are defined for
the type, and bind to the first
extension method that it finds.
And at Ms's link text :
Extension methods present no specific
security vulnerabilities. They can
never be used to impersonate existing
methods on a type, because all name
collisions are resolved in favor of
the instance or static method defined
by the type itself. Extension methods
cannot access any private data in the
extended class.
Factors that seem obvious to me would include :
I assume you would not write an extension unless you expected it be used very generally and very frequently. On the other hand : couldn't you say the same thing about sub-classing ?
Knowing we can compile them into a seperate dll, and add the compiled dll, and reference it, and then use the extensions : is "cool," but does that "balance out" the cost inherent in the compiler first having to check to see if instance methods are defined as described above. Or the cost, in case of a "name clash," of using the Static invocation methods to make sure your extension is invoked rather than the instance definition ?
How frequent use of Extensions would affect run-time performance or memory use : I have no idea.
So, I'd appreciate your thoughts, or knowing about how/when you do, or don't do, use Extensions, compared to sub-classing.
thanks, Bill
My greatest usage for them is to extend closed-off 3rd party APIs.
Most of the time, when a software developer is offering an API on Windows these days, they are leaning more and more toward .NET for that extensibility. I like to do this because I prefer to depend on my own methods that I can modify in the future and serve as a global entry point to their API, in the case that they change it.
Previously, when having to do this, and I couldn't inherit the API object because it was sealed or something, I would rely on the Adapter pattern to make my own classes that wrapped up their objects. This is a functional, but rather inelegant solution. Extension methods give you a beautiful way to add more functionality to something that you don't control.
Many other peoples' greatest usage for them is LINQ!
LINQ would not be possible without the extension methods provided to IEnumerable.
The reason why people love them is because they make code more readable.
I have noticed another MAJOR usage of extension methods (myself included) is to make code more readable, and make it appear as if the code to do something belongs where it is supposed to. It also gets rid of the dreaded "Util" static-god-class that I have seen many times over. What looks better... Util.DecimalToFraction(decimal value); or value.ToFraction();? If you're like me, the latter.
Finally, there are those who deem the "static method" as EVIL!
Many 'good programmers' will tell you that you should try to avoid static methods, especially those who use extensive unit testing. Static methods are difficult to test in some cases, but they are not evil if used properly. While extension methods ARE static... they don't look or act like it. This allows you to get those static methods out of your classes, and onto the objects that they really should be attached to.
Regarding performance..
Extension methods are no different than calling a static method, passing the object being extended as a parameter... because that is what the compiler turns it into. The great thing about that is that your code looks clean, it does what you want, and the compiler handles the dirty work for you.
I use extension methods as a way to improve the functionality for classes without increasing the complexity of the class. You can keep your classes simple, and then add your repetitive work later on as an extension.
The Min() and Max() extension methods are great examples of this. You could just as easily declare a private method that would calculate these, but an extension method provides better readability, makes the functionality available to your entire project, and didn't require making an array any more complex of an object.
Taking the sub-classing approach vs. extension methods requires a couple of things to be true
The type must be extendable (not-sealed)
All places the type is created must support a factory pattern of sorts or the other code will just create the base type.
Adding an extension method requires really nothing other than using a C# 3.0+ compiler.
But most importantly, an inheritance hierarchy should represent an is-a relationship. I don't feel that adding 1 or 2 new methods / behaviors to a class truly expressing this type of relationship. It is instead augmenting existing behavior. A wrapper class or extension method much better fits the scenario.
In some cases you can't use a subclass: string for instance is sealed. You can however still add extension methods.