And how is that any different than any language with interfaces? (Besides the implicit thing?).
Imagine a close() interface in Java. There isn't one - but having a try {...} finally { x.close(); } can be very useful sometimes. But having interface graphs made of granular interfaces makes everything slow and causes a lot of complexity when you try to think about your type hierarchy. That's why interfaces always just grow and you can't use them any longer because other classes only implement part of the api. Something usable for all of awt, swing, file handling, random foreign libraries? Unthinkable. Also, adding something in a later release (like CharSequence in 1.4) can have a wide ranging impact and requires you to change a lot of code.
In Go, you just add an interface from the union set of multiple structs api - and you can use it. No matter who wrote those structs. The value is enormous. Think of it as something like dependency injection at compile time.
https://docs.oracle.com/javase/7/docs/api/java/lang/AutoClos...
You can even leave out the finally when you use an autocloseable resource.
//r will be closed no matter what
try(Resource r = getResource()) {
} catch(SomeException e) {
}Known as structural typing and available in most modern languages.
Well, "modern" is an ill-defined concept. As far as I'm aware, structural typing is not really that common, is it? Besides OCaml and Scala, is there any relevant (used outside of academia) language that supports it?
F# also supports it, given its ML linage.
C# tricks with dynamic, although in this case it is dynamic typing, so not really the same thing.
Concepts would have fixed this, but we don't have concepts and maybe never will!
enable_if and type traits are a workaround for the time being.
Concepts lite will definitely be in the next revision.
"Duck typing is similar to but distinct from structural typing. Structural typing is a static typing system that determines type compatibility and equivalence by a type's structure, whereas duck typing is dynamic and determines type compatibility by only that part of a type's structure that is accessed during run time."
So it sounds like "duck typing" is defined in terms of run-time semantics. I guess you might be able to have genuine "compile-time duck typing" in a dependently typed language, which could be a good reason to preserve the distinction.
Otherwise, yes, they are certainly similar.
I've implemented InputStreams from byzantine transport layers in Java that work with the standard library. I don't quite understand what is special about this concept in Go (maybe it's nice for people coming from typeless, messy dynamic languages or nice languages with horribly inflexible and ad-hoc standard libraries like Python and Ruby).
I have used interfaces (or equivalent concepts) in dozens of languages and they always felt like far more of a chore, explicitly using X interface or Y interface, ugly complex declarative specs, etc. Go just makes it painless.
What people like about Go is its mix of features, not some specific one by itself.
If it had Generics that would be another thing.
However, you can wrap an interface{} using container (for example), with type-specific input and output functions, getting back compile-time type-checking. Yes, it is not optimal, and yes, you're still paying the price for type assertions.
If it had Generics that would be another thing.
I would like to see a good solution for this that integrates well with the rest of the language.
I say that because the Go designers, despite having designed a quite mediocre language, suddenly when it comes to generics the want the perfect no-compromises solution or nothing.
Say you have a type that implements the io.Reader, io.Writer, io.Closer and io.Seeker interfaces, eg. an os.File. This type would also automatically implement io.ReadCloser, io.WriteCloser, io.ReadWriter and io.ReadWriteCloser, io.ReadSeeker, io.ReadWriteSeeker, io.ReadWriteSeekerCloser, io.WriteSeeker, io.SeekCloser etc.
I read your comment, but this seems more like A sane standard library with proper interfaces and not a language feature per se (also composition to me sounds like the pattern where you create objects which are composed of other objects but that's not what you mean, right)? What you display here can be done in pretty much any language suporting inheritance (or I'm missing something), so your point is those languages don't have io.Copy out of the box?
This means that many standard types in Go implements either io.Reader and/or io.Writer. That's neat. The implicit interface implementation is definitely a language feature, and the libraries are making good use of it.
(That is not to say, you could not do the same in other languages, but you would have to specify the interface they implement rather than just adding a method with a specific layout.[0])
[0] I am not remembering the right term right now.
Close is a particularly good example. One need only look at C#'s IDisposable to see that it does, in fact, work well. A mock might noop it, another class might close an FD, and yet another might make an RPC call.
I agree that interfaces tend to have fairly narrow family tree. And, by this narrowness, there's little ambiguity about what T GetById(id int) means. As the tree expands, which happens with implicit interfaces, ambiguity is more likely. Nevertheless, there's a fairly large common vocabulary that we'd all largely agree on. Closer, Reader, Writer, Logger, etc. Even in more complex ones, I see little risk of confusion, say, http.ResponseWriter. And, something that I've noticed from Go (which I never did in C# or Java), is the tendency to favor very small interfaces, which ends up being pretty awesome.
That aside, consider that implicit interfaces allow the consumer to define the interface. For example, you create a library that has a concrete struct called MyStruct with a method called DoStuff(). You define no interface because you don't need one.
I, a consumer of your library, need an interface because in some cases I'm using your MyStruct to DoStuff and in other cases, I'm using my own implementation. So I create an interface, define DoStuff(), and BAM!, your structure now implements my interface. I don't have to change your code.
Sure, the workaround is to wrap your structure in my own which implements the interface. But how, in this case, is the implicit interface not a huge win?
Maybe, as you say, it'll screw over people who use it poorly. For everyone else, I see no drawbacks.
Haskell's typeclasses work around this by letting you define "how a type implements an interface" at either the definition of the interface or the definition of the type. (Or, strictly, anywhere else both are in scope - but that gets messy for a few reasons, so it's discouraged and GHC warns about "orphan instances" unless you tell it not to.)
That's not a problem with Go's module system. If someone is using your library, and wants to use your interface in a particular package, that's fine. If they want to use another library, with another interface of the same name in another package, that's also fine.
If they want to use both libraries in the same client package, they'll have to locally rename one or both of the imports.
Here's a container interface. It has a sort() method, which sorts the elements of the container.
Here's a different container interface. It's got a method that tells you what kind of concrete container you have, so you can understand how expensive actions like insert and sort are going to be. But the author was from England, and so called that method "sort".
You can see how that's going to (fail to) work out when you define a "sortable" interface...
Traditional (but even more contrived) examples include "draw" and "fire" doing unexpected things when applied to a gun as opposed to an image or an employee.
First, you really have to split this up into two distinct cases.
The most powerful/useful is when you create an interface in your package for a structure in a different package. It's been years since I've done .NET or Java, but I remember frequently cursing a library developer (often someone at Microsoft or Sun) for exposing a concrete type instead of an interface. In this case, I don't see any downside. Creating a:
type Render interface { Draw() }
And then using my (or someone else's) Gun structure, is no different than using the Gun structure directly and expecting it to render itself to the screen rather than come out of its holster. The interface buys you the level of indirection, but whether you use the interface or the concrete type, it's up to you to know WTF Draw() does.
This first approach is well explained at [1]
In the second case, where a library expects an interface, I agree there's a risk:
// 3rd party package package fastRender
type Drawable interface { Draw() }
func Render(d Drawable) { ... }
// my code: fastRender.Render(new(Gun))
I've lost some degree or type checking. Since my Gun's Draw doesn't explicitly implement fastRender.Drawable, maybe I'm trying to fit a square peg into a round hole. There's nothing the compiler can do to help either.
It sounds a bit like an academic concern to me. Is this something that's actually bitten people in the ass? How is it different than any method that takes primitives? I can pass a []byte of a bitmap to a Music player's Play(data []byte) method...
[1] https://medium.com/@rakyll/interface-pollution-in-go-7d58bcc...
There's also a subtle third case - I import two libraries. One of them defines Drawable, the other has something that looks like it implements Drawable. But it's just happened upon the same signature, and makes different assumptions. Maybe it even says it implements Drawable - but a different Drawable defined in a third library.
As I've said all along, I don't know how often these things are hit in practice. I think so far go development has been comparatively centralized, and so it might be seen less often than it would if it were used everywhere, but that might still be rare enough to not be a problem.
It doesn't seem like a high probability that this would happen on accident, though it might be possible to contrive an example.
I'll have to think about that.
One could possibly force the issue by including an unused function with a deliberately unique name (supports_interface_foo, or maybe a UUID or both) as part of the interface. Can functions added later fill parts of interfaces? If so, this approach would make the situation roughly that of Haskell typeclasses. If not, then this approach makes the situation roughly that of Java interfaces. In either case, it might be appropriate to some pickier interfaces.
You could include the name of the interface as part of the method name, though. But the downside of that approach is, if another interface wants to re-use the same method, things get awkward...
My point was implicit, exceptional simple interfaces are sorta magic. They are effortless, and because they are implicit, there is no harm in creating a 1 function interface (or 10 of them). Because your caller is never going to have to do an ... Implements ThingA, ThingB, ThingC, ThingD, ThingE, Thi... it implicitly supports an interface if it fulfills the signature.
So your interfaces end up being tiny (just what you need) and pervasive. This means the way your system ends up working tends to be far more aligned along interfaces than anything else.
When I speak of composition (in this case the composition of functions) I am speaking of the ability to use functions together rather freely and with little effort. This is a bit hard to explain, but it feels a bit like a modern shell, you can wire together lots of commands (functions), from lots of places that have no awareness of each other with the pipe | operator. In Go, simple, minimalistic interfaces act as the glue and let you compose lots of diverse functions together exceptionally quickly.
I'd say it's a better C, not a better C++ (or Java/C#)
For me, most modern uses of C fall into 3 camps:
+ extreme portability
+ complete, low level resource control
+ foundational libraries
And Go doesn't really qualify for any of these. For the latter 2 largely because of the GC.
I think in practice, that its engineering bent i.e. no frills language and high quality tooling, will more likely see it being used in the server-side application/middleware space. I guess not surprisingly.
So it really is up against Java, Python, Ruby and C++ and is, therefore, interesting in that it isn't trying to compete on lingustic goodies.
Yeah, I agree with your points. Which is a shame really.
io.Copy(dst io.Writer, src io.Reader)
It reads data from reader and writes to writer... simple. Now what makes this little function so darn useful is it takes anything fulfilling its interfaces (io.Writer and io.Reader). The first way you will probably use it will be to copy between some stream and a file without having to eat up all the memory to store the buffer (not using ioutil.ReadAll for example)... but then you realize you can use a gzip compressor on the writer side, or a network socket, or your own code... and io.Copy works with anything that fulfills its interface.
As you build out a complex application, you start by creating your own functions that take advantage of existing interfaces foo.OCR(dst io.Writer, src img.Image). After that you start building out your own interfaces... like a MultiImage interface that has ImageCount and ReadImage methods that returns the count of images and binary data... but then you realize the images could be big, so you make the ReadImage method return an io.Reader... and now you have gone full circle and are using io.Copy to copy imageX from a stack of images to return to your OCR function that will output the data to an io.Writer which you made actually a gzip writer because text compresses well.
Beyond composition -- obviously, the concurrency and messaging is nice and when you need it vital. The other thing that will help make Go "click" is being very "data oriented" in your design... be vicious and minimal: http://youtu.be/rX0ItVEVjHc (great talk on data oriented design) and be absolutely pragmatic, focus on getting shit done, always...