people working actively in "modern" Go codebases with generics, is it becoming like a Java/Spring kind of codebase?
I understand why you might need generics, but I hate how Java ecosystem exploits it so much that, reading code becomes so difficult, because it was inherited from 10 levels of parent classes and in some cases Java Beans get created based on generic types and their parent classes are abstract classes.
Neither Go nor Java's generics is rich enough for the problems you describe. They give you the means to turn IntList into List<I>, but not into <L>Int or <L><I>.
So I doubt the famous AbstractProxyBeanInterceptors are coming from an overuse of generics.
I remember my coworkers having managed to write Java in pre-generics Go. The code had functions wrapped in functions wrapped in yet more functions, and interfaces only ever implemented by one type to replace the generics abuse. Life finds a way.
Yeah, Go people were always been perfectly capable of creating abominational messes before generics arrived. Once you've looked at an interface referencing an interface referencing an interface referencing an interface which is only ever instantiated by one (1) type, you realise that all hope has been lost and your only recourse is blatant sarcasm, pointed replies on Slack[0] and PRs, and finding the perpetrators after dark with a big sack of flour.
[0] Other corporate messaging systems are available.
"people working actively in "modern" Go codebases with generics, is it becoming like a Java/Spring kind of codebase?"
I have yet to encounter a 3rd party library that uses generics that isn't a data structure library of some sort. The only person I have seen doing crazy things with generics is... me. In my own code, for the most part [1]. In general I don't even see generics that are so much as parameterized by a meaningful interface like "type BlahBlah[R io.Reader]"; generics are so often parameterized by "any" that a casual reader might be forgiven for thinking that that is all a Go generic can do. As is often the case, the most complicated generic type signatures are the one in the standard library and even they aren't really all that bad, it's really just about the "~" operator allowing support for "all types that are either 'string' or something of the form 'type X string'".
I see the occasional person complain about how generics have ruined Go by making them too complicated. I push back on them with basically the previous paragraph. None of them have ever replied to press their point any farther. I interpret this to mean that their experience with the feature is in fact the same. At this point if you're buried in super complicated generics it is almost certainly someone on your own team and you should tell them to stop.
Similarly for iterators, by the way. The way Go did iterators is a bit odd to implement them yourself. Won't deny that. But it does compile down very efficiently, and only matters to the people writing the iterator not the consumer. Anyone who claims it has wrecked the language is invited to explain to me why it is that they are writing so many dozens of iterators all the time first, because as near as I can tell they aren't getting that from 3rd party libraries or the standard library.
[1]: The exception being https://github.com/thejerf/mtmap if you want to see something that I did find a use for, and you may have a use for, but is not something I would suggest using all the time. Very specific use case I had.
I’ve found generics useful at handling database-backed objects at the write boundary. Since the layout of those objects is generally pretty important, having middleware-ish things that handle interface/supertypes but return generic concrete types saves on a lot of reflection and casting.
I think Go is more commonly used in modern microservice architectures rather than in older monolithic architectures where the kind of Java code you're describing happens. Might have some impact on the level of complexity that can get out of hand.
In the Go codebases I have worked on, generics were not overused IMHO. It did allow us to get rid of some type specific helper functions and made collection based operations like sorting a lot nicer to work with.
Rust has had generics for 10 years and doesn't have this problem. I think it's more a culture problem in Java than anything (and arguably a historical one).
I haven't seen that in the large codebase at work, for what it's worth. The thing that's surprised me about generics in Go is how rarely they're needed in practice. Very nice to have in the right place, but the vast majority of our source files still get along fine with plain old typed maps and slices.
I have been a full-time go developer since 2015 and I have quite literally never written a line of generic go code. I can't even recall seeing any written at any {job} since they've been introduced...
> Go’s interface system works at runtime. The specific type of a value passed into an interface parameter is resolved while the program runs. This dynamic dispatch clashes with generics being resolved at compile time.
What if the compiler generated both vtable and monomorphized variants for each method? Most calls would use the static version but places that need it like interface generics call the virtual method.
The problem is that your vtable would have to have a monomorphised copy of the generic method for each possible instance, this requires whole program analysis just to generate the interface — and even then I would not be surprised if you could generate calls which the compiler was not able to relate to the interface — and then you get to the potential size explosion of your vtable as it has to hold an unbounded number of methods.
Note that despite having much stronger static semantics and much weaker reflection than Go Rust has the exact same limitation: a trait method with a non-lifetime generic parameter is not dyn-compatible. C++ is the same, a template member function (generic method) can’t be virtual (dynamically dispatched).
There's a little more nuance to this. The compiler knows the static type of any value placed into an interface value, and emits code to place a concrete type pointer (or itab pointer) into the interface value. You can see this in the disassembly.
Later when you pass this value around and assign it to places, sure, there may be a different type depending on where the interface value came from. But the construction of these values is very much static.
In this sense I wouldn't really call this "resolv[ing] while the program runs" unless you take a very broad sense of "resolution."
The construction of interface values is a very shallow mean and essentially irrelevant to the need for interfaces.
Whereas the runtime resolution of operations performed through an interface (dynamic dispatch) is the primary purpose of and use care for interfaces, with the dynamic resolution of wrapped types (type erasure / downcasts / type switches) is the secondary one.
Go is often thought of as a successor of C. C doesn't have methods, only global functions. From my perspective, Go added methods primarily so that you can use them in combination with interfaces. Given that interfaces don't support generic methods, I'm personally not convinced that this feature was worth adding.
> From my perspective, Go added methods primarily so that you can use them in combination with interfaces.
They also give you a limited form of overloading. Without methods or overloading, you end up in the situation that C and Scheme are in where every operation on a data structure has to redundantly have the data structure in its name like:
In Go you would methods for that: my_list.Clear(), my_queue.Clear() and my_map.Clear(). Now you can define a Clearer interface, which has only the Clear method. That allows you to write a function clearAndLog(item Clearer) and it will work with the list, queue and map.
The methods in an interface can be implemented by many different types, and it's often hard or impossible to determine which of those types will be passed in to a function. For example, the io.Reader interface is implemented by many different stream-like types, and functions which accept io.Reader arguments generally can't make assumptions about which of those types they'll get.
I understand why you might need generics, but I hate how Java ecosystem exploits it so much that, reading code becomes so difficult, because it was inherited from 10 levels of parent classes and in some cases Java Beans get created based on generic types and their parent classes are abstract classes.
So I doubt the famous AbstractProxyBeanInterceptors are coming from an overuse of generics.
[0] Other corporate messaging systems are available.
I have found Spring like magic with pre-processor macros, sed and awk, in the glory days of Yourdon Structured Method and enterprisy C code.
This is not (yet) a thing in Go. Go does not have classes and inheritance. Interface is more like duck-typing.
I have yet to encounter a 3rd party library that uses generics that isn't a data structure library of some sort. The only person I have seen doing crazy things with generics is... me. In my own code, for the most part [1]. In general I don't even see generics that are so much as parameterized by a meaningful interface like "type BlahBlah[R io.Reader]"; generics are so often parameterized by "any" that a casual reader might be forgiven for thinking that that is all a Go generic can do. As is often the case, the most complicated generic type signatures are the one in the standard library and even they aren't really all that bad, it's really just about the "~" operator allowing support for "all types that are either 'string' or something of the form 'type X string'".
I see the occasional person complain about how generics have ruined Go by making them too complicated. I push back on them with basically the previous paragraph. None of them have ever replied to press their point any farther. I interpret this to mean that their experience with the feature is in fact the same. At this point if you're buried in super complicated generics it is almost certainly someone on your own team and you should tell them to stop.
Similarly for iterators, by the way. The way Go did iterators is a bit odd to implement them yourself. Won't deny that. But it does compile down very efficiently, and only matters to the people writing the iterator not the consumer. Anyone who claims it has wrecked the language is invited to explain to me why it is that they are writing so many dozens of iterators all the time first, because as near as I can tell they aren't getting that from 3rd party libraries or the standard library.
[1]: The exception being https://github.com/thejerf/mtmap if you want to see something that I did find a use for, and you may have a use for, but is not something I would suggest using all the time. Very specific use case I had.
In the Go codebases I have worked on, generics were not overused IMHO. It did allow us to get rid of some type specific helper functions and made collection based operations like sorting a lot nicer to work with.
What if the compiler generated both vtable and monomorphized variants for each method? Most calls would use the static version but places that need it like interface generics call the virtual method.
Note that despite having much stronger static semantics and much weaker reflection than Go Rust has the exact same limitation: a trait method with a non-lifetime generic parameter is not dyn-compatible. C++ is the same, a template member function (generic method) can’t be virtual (dynamically dispatched).
> However, Go’s interface system works at runtime. The specific type of a value passed into an interface parameter is resolved while the program runs.
Later when you pass this value around and assign it to places, sure, there may be a different type depending on where the interface value came from. But the construction of these values is very much static.
In this sense I wouldn't really call this "resolv[ing] while the program runs" unless you take a very broad sense of "resolution."
Whereas the runtime resolution of operations performed through an interface (dynamic dispatch) is the primary purpose of and use care for interfaces, with the dynamic resolution of wrapped types (type erasure / downcasts / type switches) is the secondary one.
They also give you a limited form of overloading. Without methods or overloading, you end up in the situation that C and Scheme are in where every operation on a data structure has to redundantly have the data structure in its name like:
SICP: 2.5 Systems with Generic Operations
https://sarabander.github.io/sicp/html/2_002e5.xhtml
Well in C at least we now have this:
...although it turns out the other nice thing about methods is automatic namespacing.And everyone gets to invent their own vtable implementation since the 1980's.