Go: Support for Generic Methods | Hacker News Reader

Go: Support for Generic Methods

github.com

311 pointsf311a274 comments

> Go doesn't support such generic interface methods because we don't know how to implement (calls of) them, or at least we don't know how to implement them efficiently.

I don't really understand this argument. I read the discussion linked to[1], and yeah, monomorphization approaches (whether at compile time, link time, or runtime with JIT) are obviously going to be difficult or impossible, but the reason against using runtime reflection is mostly that it's slow. But that runtime reflection is exactly how you would work around it today.

For the Identity example, could the interface be compiled to be basically equivalent to:

Identity(any) any

and then at the callsite add a cast of the return type to T?

I suppose primative non-pointer types add a bit of a wrinkle but even if it generic methods was restricted to pointer types, that's better than nothing. And the number of those types is relatively small, so when the implementation is compiled it could just instantiate method implementations for all the primative types, if they apply, and then maybe remove them if they aren't needed at link time.

Of course it's also possible there is some detail I've missed.

[1]: https://go.googlesource.com/proposal/+/refs/heads/master/des...

> but the reason against using runtime reflection is mostly that it's slow.

More specifically, it is that it would introduce surprising performance cliffs – code becoming surprisingly slow due to seemingly unrelated changes.

Though BTQH I think an even more important argument is that you would need to have effectively two generics implementations, one working at runtime and one working at compile time. That's a lot of complexity, with surprising failure modes if these two are not bug-compatible.

> But that runtime reflection is exactly how you would work around it today.

I think the overwhelming majority of people will "work around it" by just not trying to use generic methods.

> you would need to have effectively two generics implementations, one working at runtime and one working at compile time

My understanding is that go already has a hybrid system works at compiletime and sometimes at runtime.

I'm not sure what you mean. Perhaps you are referring to the reflect package? In that case, yes, that exists. But it is limited in its power (for example, it doesn't allow to create types with methods – precisely because of the difficulties we are talking about) and a comparatively frequent source of bugs. If anything, it provides pretty strong evidence for the problems with this approach.

https://go.dev/doc/faq#generics_implementation

My point is for interface generics it could just always use a single instantiation. Similar to what java does.

Or alternatively, go could go the other direction and add a new type of interface that is only for use in generic constraints, and then generic methods could be part of that interface, but not normal interfaces, so that the generic methods could be called from other generic functions. That would be similar to rust and c++.

When I said "you need a runtime implementation of generics" I did not mean "you need dynamic dispatch". I meant "you need a type-checker and at least limited code generator at runtime".

> Similar to what java does.

Yes, if you completely forego the static implementation, you can deal with just one. Then all code is slow.

Java can get away with it, because it has a JIT compiler – so it isn't subject to the same "no runtime code generation" design restriction Go is. That also means, it can run on fewer platforms than Go.

> add a new type of interface that is only for use in generic constraints

Yes, that would be feasible. It has been ruled out, because we really don't want to have interfaces that can not be used as types. They currently exist, but the goal is to remove that restriction, not to add more of the same.

I'll note that either way, at that point we are firmly outside the realm of this particular thread. It's no longer what I was responding to.

Sorta, but it's important for the calling convention that the compiler is consistent on what is done at compiletime vs runtime. Because methods are "normal functions" for the calling convention (and can be assigned to function-typed variables), there would be a lot of gymnastics required for the compiler to make runtime-generated variants of methods work.

BTQH?

Maybe a typo for To Be Quite Honest?

> Of course it's also possible there is some detail I've missed.

Can't speak too deeply for Go specifically, but I do know on .NET one of the big reasons generic methods where T is a structure gets monomorphized per type, is so that stack size is adjusted and potentially even arg passing (i.e. large struct) as far as the caller/callee.

Like Java‘s generic erasure? Primitive types aren’t supported at all, have to used boxed ones. Its slightly annoying but not too much. You can fall back to arrays to have performant primitive containers.

The real problem is that Go produces interface implementations dynamically.

The determination wether type T implements interface I is made at runtime. So is generation of the necessary vtables to produce the interface implementation.

So you can do things like this in package a:

  type S struct {
      //...
  }

  func (s \* S) Foo() {
      //..
  }

and something like this in package b:

  type Foo interface {
      Foo()
  }

  func DoSomethingWithAFoo(f Foo) {
  }

and something like this in package c:

  func Stuff(obj any) {
      theFoo, _ := obj.(b.Foo)
      theFoo.Foo()
  }

And then do:

  var s a.S
  c.Stuff(s)

And everything works.

For generic functions, go uses a strategy similar to C++ templates: when you call a generic function the compiler statically produces a concrete specialization of the generic function based on the inferred types for generic parameters.

That is, if you do:

  func Bar[T any](x T) {
    //...
  }

And you do:

  var x int
  var y string
  var z float64

  Bar(x)
  Bar(y)
  Bar(z)

The compiler statically generates 3 versions of Bar, one that takes an int, one that takes a string, and another that takes a float64.

These two things don't work well together. If I have a variable typed as `any`, and I want to cast that to an interface, I need to dynamically determine 2 things:

1. The shape of the interface's vtable. The go runtime does this by iterating over the runtime metadata for the interface type.

2. For each named method in the interface's vtable, the address of the concrete function to stick in that vtable slot. This is done by accessing the reflection metadata for the implementing type. It verfies the method with name X for type T matches the required signature for the method with name X for interface I, then sticks that method pointer into the appropriate vtable slot.

The problem, however, is what happens when the method with name X is generic. There may, or may not, be an actual concrete method for the set of type parameters. It's possible that statically type T does implement interface I (via generic methods) but that dynamically it doesn't because the particular generic instantiation needed for the particular interface was never made statically.

Prior to go 1.27, this was never an issue, because methods could not declare their own type parameters. They could reference the generic parameters of the receiver, but once the receiver type was known, there was only ever one concrete method X for that receiver.

Once you allow methods to have their own generic type parameters, the compiler can introduced several different concrete implementations for a method X.

This is ok, when you do somethnig like:

  var x SomethingWithGenericMethods
  x.Foo(1)
  x.Foo("hello")
  x.Foo(1.2)

Because the compiler knows statically from the Foo call sites which concrete methods it needs to generate.

But, when you introduce a dynamic cast:

  var x SomethingWithGenericMethods
  var i SomeInterface

  i = x.(any).(SomeInterface)
  i.Foo(1)
  i.Foo("Hello")
  i.Foo(1.2)

It's entirely possible that the necessary Foo implementations don't actually exist in the binary.

So, go 1.27 introduces generic methods, but it gets around this problem by saying:

1. Interface types can't define generic methods

2. Generic methods can't be used to implement interfaces

Thus, it allows adding generic methods without introducing the issues that crop up with dynamic interface implementations.

This will finally let me make the monad library I've been dreaming of for years. Be afraid.

A monad library in go can really on have one name… …

Go nad or go home

Underrated

Strife

Weeee

GoMad

Mongo

(is appalled!)

Mo-go?

It's (sadly) still not possible to express monads with this change, since generic methods can't implement interfaces. You'd probably want something like:

    type Monad[T any] interface {
        Bind[U any](func(T) Monad[U])
    }

However this requires the Bind method to be generic, which still isn't allowed in an interface

I am not very familiar with Go and especially not its generics support. Can you implement the "join" version instead of the "bind" version, where you turn a T[T[a]] into a T[a]?

Hmm I wasn't familiar with join, but it looks like you still need join + fmap for the construction? I believe fmap would also need a generic method

Yeah you would, that's true.

Funnily enough, Go's generics were designed by the same guy who introduced monads to computer science. Everything comes fill circle.

> Funnily enough, Go's generics were designed by the same guy who introduced monads to computer science.

No contributor to Go is responsible for "introducing monads to computer science", as the Monad concept is a member of (or defined by if you prefer) Category Theory[0].

0 - https://en.wikipedia.org/wiki/Category_theory

As you point out, monads come from category theory, not native to computer science. Thus there had to be someone to introduce approaches to applying monads in computer science. The paper usually credited with that is: https://homepages.inf.ed.ac.uk/wadler/papers/marktoberdorf/b... Which the parent rightfully points out was written by the same person primarily responsible for the design of generics in Go: https://homepages.inf.ed.ac.uk/wadler/topics/go.html

> As you point out, monads come from category theory, not native to computer science. Thus there had to be someone to introduce approaches to applying monads in computer science.

With full respect given to Wadler and his contributions to computer science, the very paper you cite authored by Wadler declares:

  The concept of a monad, which arises from category theory, 
  has been applied by Moggi to structure the denotational 
  semantics of programming languages.

Therefore, one cannot assert that a computer scientist whom identifies a predecessor's contributions such as above is, in fact, responsible for said contributions.

And with full respect also given to Moggi, introductions are only introductions if they are heard.

If we were talking about the first person to discover monads, Wadler clearly would not be he. Wadler was but a wee toddler barely sputtering out his first words when the term "monad" was coined. As we are talking about who introduced monads to computer science, the signs I see continue to point to Wadler. His work is, by all appearances, what caught the attention of the computer science community. "Monads for functional programming" is regularly cited as the seminal paper. They are strongly associated with Haskell. It would even appear from the previous comment that you only came to learn about Moggi because of Wadler making his introduction, which echos Moggi not being particularly influential socially.

If you have evidence to suggest that Moggi played a bigger role in introducing (not inventing) the concepts, I am definitely keen to learn about it.

> If we were talking about the first person to discover monads, Wadler clearly would not be he.

We are not. Remember what the OP stated, to which I originally replied:

  Funnily enough, Go's generics were designed by the same guy 
  who introduced monads to computer science.

This is demonstrably incorrect, as Wadler did not "introduce monads to computer science" and is what I disagreed with alone.

Did Wadler take monads to new and important levels in both computer science and software engineering? Absolutely!

> It would even appear from the previous comment that you only came to learn about Moggi because of Wadler making his introduction, which echos Moggi not being particularly influential socially.

I quoted the Moggi attribution due to it being present in the paper you kindly provided. Whether Moggi has any influence socially is irrelevant as it pertains to the original thesis discussed above.

> If you have evidence to suggest that Moggi played a bigger role in introducing (not inventing) the concepts, I am definitely keen to learn about it.

Again, this is not germane to the aforementioned OP statement.

Moggi's social influence is everything as introductions can only happen in the social realm. It continues that all evidence I can find suggests that his work went unnoticed until Wadler pushed it into view. In other words, people came to learn about his work only because of Wadler making the introduction. Nobody, except where one may find it trying to be implied in https://news.ycombinator.com/item?id=48304692, has suggested Wadler invented monads. The OP clearly wrote "introduced", not "invented".

On that note, calculus come from physics. AI should really hand back all those jacobians and hessians.

We already have monads at home (return X, err)

Can we have Exception monads? Asking for friend.

> Can we have Exception monads? Asking for friend.

This is nonsensical. Monads define a strict set of behaviors formalized as "monad laws"[0].

Perhaps what you want is a container which adheres to monad laws capable of abstracting exceptions. Two exemplars of same are Haskell's Data.Either[1] and Scala's Either[2].

0 - https://wiki.haskell.org/Monad_laws

1 - https://hackage-content.haskell.org/package/base-4.22.0.0/do...

2 - https://www.scala-lang.org/api/3.8.3/scala/util/Either.html

I don't think it's nonsensical, it's just another name for the same thing. E.g. in the Haskell wiki it says, "the Error monad, also called the Exception monad".

https://wiki.haskell.org/index.php?title=All_About_Monads

I was unaware of "Exception monad" being an industry equivalent term for Either/Error. Given no other context, desiring an "Exception monad" could be interpreted as "a type which handles raising Exception types." Which is how I did.

Thank you for clarifying this for me.

> Perhaps what you want is a container which adheres to monad laws capable of abstracting exceptions

That is what I meant. Struggling to picture what the other "nonsensical" thing is.

As I commented here[0], my interpretation of:

  Can we have Exception monads? Asking for friend.

Was a desire to have "a type which handles raising Exception types." Until now, I was unaware of "Exception monad" being an alternate moniker for Either/Error types.

0 - https://news.ycombinator.com/item?id=48318468

Go becoming a proper 21st century language, is like pulling teeth.

It is Apple's school of design, think different, ah, actually, there are reasons why the fence is in the middle of nowhere.

Then the design ends up half way there versus being done properly from the beginning.

Your comment could have been, "I've never agreed with the design philosophy behind Go." I've always appreciated the apparent Go design philosophy and feel like it most matches my lessons learned from 20 years in software. Feature minimalism is a feature for languages targeting organizations with thousands of programmers. If by 21st century language, you mean one that has become unrecognizable through multiple generations of fashionable feature and ecosystem thrash then I'm all for Go not becoming a 21st century language. Language should be boring if the target environment is large teams with varied skill sets. Plain speak. Low jumpy behavior. No cream, no syntactic sugar. Get the job done.

“Done properly from the beginning” means explaining why a particular feature is either included or not. In this sense, Go is done properly from the beginning. It would be wrong to add every popular feature uncritically.

"They are likely the two most difficult parts of any design for parametric polymorphism. In retrospect, we were biased too much by experience with C++ without concepts and Java generics. We would have been well-served to spend more time with CLU and C++ concepts earlier."

Yeah very critically.

You can’t be omniscient, I think.

It is sufficient to actually care about history of programming languages design, acknowledge the paths trailed before since FORTRAN came up in 1958, no need for omnisciency.

Less "we know better", more "actual history".

Lack of generic methods was really surprising to me when I was first trying to use generics in Go. Nice to see it being actually implemented

To be replaced by the surprise when you figure out these methods don't implement interfaces.

Still, in this case, half the feature is better than none at all, IMO.

Generic interfaces are going to be implemented later too if I'm reading correctly. So no real surprises there :). I guess the only surprise yet is that generic interfaces aren't supported, so generic methods physically can't satisfy any interface

Generic interfaces already exist. Generic interface methods, which would be relevant, are not planned. The reason is outlined early in the proposal: nobody knows how to implement them efficiently. Rust has the same problem, for what it's worth: dispatchable functions on dyn-compatible traits cannot be generic [1].

[1]: https://doc.rust-lang.org/reference/items/traits.html#r-item...

Or to look at that from another angle, if you were to define a Trait which has generic methods that Trait won't be "dyn-compatible" meaning that you can't do dynamic dispatch with this trait, which may be irrelevant to you (if you don't want dynamic dispatch anyway) or a showstopper (if you needed it, now your project won't compile).

That is another way of looking at it, but given the topic, you're gonna have to expand or contextualise that. I Rust a fair bit, and only barely follow.

Actually in hindsight I think my perspective was less helpful because you can write a dyn compatible trait with a generic method it's just that you can't call the method via the trait objects, dynamic dispatch isn't possible for your function. So the original way to think about it was superior.

FWIW I found, so far, that bringing up dyn-compatibility to Rust people was very useful in helping them understand why Go's interfaces won't ever have generic methods.

The one additional piece of information you need is that in Go, all interfaces are supposed to be trait objects. The exception are union-elements, but that's really a restriction the Go team is trying to remove, not a model to base more features on.

Is it because of the kinda built-in duck typing, for want of a better word? The thing where if you have a method (or methods) that matches the signature of method(s) of an interface, you implement the interface without explicitly declaring so?

> for want of a better word

Structural typing is the term typically used to describe "static duck typing".

I’d describe structural typing as a special case of duck typing FWIW.

I suppose. That special case is that it is evaluated statically, whereas duck typing is evaluated dynamically. That's the whole difference between them. They are otherwise identical concepts.

But who is to say that dynamic evaluation isn't the special case?

More specifically, it is because of interface type assertions – the fact that if you have a value of some interface type (e.g. `any`), you can dynamically assert that it is another interface type (e.g. `io.Reader`). A good example of that is `io.Copy`: https://cs.opensource.google/go/go/+/refs/tags/go1.26.3:src/...

This aspect is what prevents you from statically knowing which interface-implementations you need to generate for a specific concrete type. There could always be new ones added at runtime.

I didn't see anything beyond "this doesn't prevent us from doing it" yet.

Did you?

slowly implementing all the things they said we didn't need

Watching Go's development is like reliving the development of Java (which also didn't have generics at first), but over decades instead of years. Cannot wait for Go to implement an error handling system in the 2030s.

Its very funny watching certain segments of the programming industry rediscovering incredibly basic programming principles, after railing against them for so long. The AI people are starting to try and create formal specs to force the AI to generate an exact output, which is absolutely hilarious to me

Dynamically typed/untyped languages finding that strict and visible typing is actually good is another

I feel like there was a name for formal specs that produce an exact output, but it escapes me…

Another one, WebAssembly people recreating application servers, and networking object models.

Thankfully we already have Java, so using Go is really for the scenarios where it cannot be avoided.

Thankfully some of us have taste.

Which is why we use platforms and languages that offer superior GCs, optimization, speed, and observability, like the JVM and Java.

They didn't say they never wanted to do generics, but that they did want to take their time and do them right.

Debatable how much they have been "right", although this gets them somewhat closer. And I think they have not been "wrong" in the ways they wanted to avoid (they referenced some issues with Java generics as prior art, although I forget the details).

From another commenter here:

> The post quotes the Go FAQ as saying, "we do not anticipate that Go will ever add generic methods".

> “What do you think? There was a man who had two sons. He went to the first and said, ‘Son, go and work today in the vineyard.’ “ ‘I will not,’ he answered, but later he changed his mind and went. “Then the father went to the other son and said the same thing. He answered, ‘I will, sir,’ but he did not go. “Which of the two did what his father wanted?” “The first,” they answered. Jesus said to them, “Truly I tell you, the tax collectors and the prostitutes and the Go language maintainers are entering the kingdom of God ahead of you.

https://www.bible.com/bible/compare/MAT.21.28-31

Whether they made a good decision or not, or changed their mind or not, is beside the point (I think they did, and they did!). I was merely replying to the claim that

> They didn't say they never wanted to do generics, but that they did want to take their time and do them right.

Is anyone actually mad about this, or do people just bring it up to stir the pot? Who cares what the FAQ says? They've worked out a way to add it easily in a backwards compatible way that can solve some problems. They had not identified this solution at the time they wrote the FAQ, and Go has been perfectly usable without this feature for 16 years.

I'm not mad, I'm a proponent of stronger type systems. I'm just correcting the record about

> They didn't say they never wanted to do generics, but that they did want to take their time and do them right.

What’s the correction? The two claims are not in conflict. Saying “we don’t expect to ever add X” is not equivalent to “we never wanted to add X.” It simply means that they didn’t think it would happen, which can coexist with an underlying willingness to consider it if a suitable approach appeared.

You added the word "want", the OP said "need". "We don't ever expect to add X" implies "we don't think we need X."

Clearly we don’t need this feature. Just because the Go team decides to implement a feature doesn’t imply that they must think that the language needs the feature. You’re searching for contradictions where none exist.

Ah, then they’re implementing needless features! What’s worse! lol

Most programming language features are not strictly needed. They’re just quality of life improvements that are on balance a good addition to the language.

Sure, but if you take that view, everything above assembly is a QOL improvement and we really don't need anything. That definition is pretty useless though.

Hmm, I don't think so. Go needs a string type to be a usable high-level programming language, but it doesn't need, say, the 'any' alias for 'interface{}'. Some features are more central than others. Perhaps none are 100% essential or inessential, but some features are clearly closer to one end of the spectrum than the other.

We care that time and time again, when anyone ever brings up a criticism of the language, they’re told that everything is just fine and it’s not a problem and we just don’t get the Go Philosophy. There’s not a problem, stop trying to make Go like every other language, and changing things would make the language more complicated and worse.

Then when the language is inevitably changed for the better, resolving the complaint, suddenly it was always going to happen and it was just a matter of getting the details right.

Every other language community I can think of is more than willing to acknowledge the shortcomings of their language. “Yeah, this kind of sucks in principle but it’s not something that gets in the way in practice” is a fine perspective. So is “this was a tradeoff; we went in this direction and these are the resulting downsides”. But the golang community practically trips over themselves to constantly argue that obvious shortcomings in the language are actually a good thing and we just don’t get it.

Nobody is saying the language shouldn’t improve. We’ve all been begging the language to improve. But we’re also tired of the constant, obvious, and shameless gaslighting from the community whenever things do get better. You aren’t going to like the comparison, but it’s extremely Trumpian.

Yeah, I worked with a guy in the late 2010s - one of the most painful people I've ever worked with - who would tell anyone that would listen that Go (as it was in 2018) was the perfect programming language - it had all the features you'd ever need - no more, no less. It doesn't need generics, the package management story is fine etc. Thankfully he's been out of my life for a long time now but I believe he's still writing Go, and I bet that he's telling anyone that will listen that Go (as it is in 2026) is the perfect programming language and that its implementation of generics was necessary and perfect etc.

He wasn't the only one but he certainly took it to the extreme.

This is an outlier. The Go team and community never endorsed that. In fact, their position has always been the opposite. To give just one example, see [1].

[1]: https://research.swtch.com/dogma

I think it’s pretty clear this post was a response to the clear dogma within the community.

> But we need help from everyone. Remember that none of the decisions in Go are infallible; they’re just our best attempts at the time we made them, not wisdom received on stone tablets.

You just can never trust what the go team says.

Yeah, because software is not meant to evolve and people are not meant to either.

You're not meant to gaslight about the evolution of software.

You're not meant to apply poor man's psychology everywhere just because you have heard about a certain psychological term.

It's a perfectly good term for the overall and historical attitude of the go dev team.

Changing your mind is not gaslighting, people just change their mind sometimes.

It's not this one thing, it's that yhey consistently do this

But they haven’t added generic methods, really. This change just lets you use method syntax in cases where you could have used a function. It’s a pretty conservative change that I think a lot of people here are misunderstanding. Actual generic methods wouldn’t really make sense (for the same reason that you have similar restrictions on dynamic trait implementations in Rust).

There's a critical difference between "we don't think we will add it" (what you quoted) and "we won't add it because we think you don't need it" (what OP claims)

Depends you who "they" is. If you mean the go development team, then yes, they said they wanted to "take their time and do them right"¹. But there are many "gophers" who did say that there was no need for generics, and that it shouldn't be added to the language.

¹ I would argue that it is really, really hard to add generics to a language after it has already matured, and still "do it right" than to add it in the beginning. At least if you care about backwards compatibility. Backwards compatibility adds a lot of constraints to your generics system that will almost certainly lead to a sub-optimal design. And you will be stuck with a standard library, and a lot of existing ecosystem code that would benefit from generics, but don't because generics didn't exist when they were written. This is a lesson I wish go had learned from Java's generics.

I agree that they were clearly not in a hurry. I disagree that they are doing everything right. I am interested to see how they will fix the 'million dollars mistake'.

It's not a bad thing to realize that one can be wrong and then strive for change.

Maybe, but personally I've become quite tired of programming languages "organically grown" as opposed to properly designed the first time. After a good decade of C then C++, I found ANSI CL (despite being a massive compromise and unfinished) much more coherent and complete than both.

I know Go is justly criticized for many of its design decisions, but it still feels well-designed and "small" to me in day to day usage when many other languages don't.

Eh, the thing with generics coming late is pretty much what I meant by "organically grown".

My best litmus test these days is support for multidimensional arrays because it's always needed at some point in general purpose languages. CL and Ada had it right from the start while C++ needed C++23/26 to get std::mdspan and we still need to wrap it to pass the underlying/owned memory pool around (https://rosettacode.org/wiki/Multi-dimensional_array for more).

Doesn't every language support multidimensional arrays? It's just an array of arrays, no? What am I missing?

An array of arrays is an extremely inefficient and error-prone way to represent multidimensional arrays.

If I want a 1000x1000 array, representing it physically as a single 1000000-element array requires one allocation, and processing it element-by-element (assuming it's stored in the same order we're iterating over it) is sequential in memory and therefore very efficient.

Representing it as 1000 separate 1000-element arrays requires 1000 allocations, and pointer-chasing every time we move from one row to the next.

Isn't an array of arrays by definition the sequential implementation?

Otherwise you would have an array of pointers to arrays. The usage (syntax) for them would be the same but the performance would not be.

They also have different uses. You would expect an array of arrays to be an array of arrays which share the same length. For an array of pointers to an array you would expect dynamic length arrays contained within the original array.

Even in c++ could you not just define some int [1000][1000]foo? I've never really used C++ but my C knowledge assumption is that is 1000000 continuous elements.

The C++ way to do it currently would be:

    std::array<std::array<T, N>, M> data;

Which is contiguous

    int data[M][N];

also works fine and is contiguous in C++

Edit:

For the stack at least. On the heap, you'd need to use a single std::vector<int> and do the indices manually, or use mdspan

I does not work fine in C++ when N and M are not compile-time constants, which is basically always the case in any interesting numerical algorithm. Also not in Rust.

It works fine in C though, or FORTRAN, or Ada, or ALGOL 60, ...

Which is why std::mdspan exists, and std::linalg.

NVidia has pivoted to design CUDA hardware with focus on C++ back in , and seems to be doing quite well for them.

CppCon 2017: "Designing (New) C++ Hardware”

https://www.youtube.com/watch?v=86seb-iZCnI

They were also the ones sponsoring the ISO work on mdspan, while HPC research labs are pushing for linalg on top.

I would rather be using Ada today, but that isn't how the world moves.

I see that they spend time making their hardware run general software, but I can't see anything specific in GPU hardware to std::mdspan.

I respect Ada but I would not want to use it. But I have a choice between C++ with hmdspan and C99's arrays, I choose C99 any time.

Why is that? I find Ada much nicer than the C-languages when it comes to arrays: A'Range, A'Length, A'First, and A'Last are super-useful, as is the unconstrained array.

You can even use unconstrained arrays to provide the same functionality that Optional does in functional-programming, provided the element-type can be an element of an array:

    -- Here we define an index-type with one value.
    Subtype Boolean_Index is Boolean range True..True;
    -- And here we define an array indexed by it, but can also have length 0.
    Type Optional(Boolean_Index range <>) of Element;

And there you have the mechanism for Optional; just use "For Object of Optional_Array Loop" to enclose your operations and bam, it works perfectly.

I guess you aren't their target customer anyway, NVidia isn't that found of pure C code, with first class tooling for C, C++, Fortran, Python JIT, Ada and most recently Rust.

The std:mdspan proposal came from NVidia employees, alongside AMD and HPC research labs.

https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2022/p00...

Yeah, I remember discussions on comp.lang.c calling programming with Ada, or even Modula-2, programming with straightjacket.

Meanwhile governments and national security bodies got another point of view.

You mean the C99 arrays Google paid the work to clean from the Linux kernel?

The thing is that I rewrite high performance numerical code on GPUs and the CUDA part is what sucks most. And the moment one uses templates, the compilation times make it insufferable. I really do not understand why people put up with this garbage. I am really looking forward to the day where I can remove CUDA from my projects and replace it with compiler-supported offloading is really

The kernel removed VLAs, I am more talking about vm types. But even for VLA - while I had a small role in that undertaking myself - I think it was a stupid mistake from a security point of view to remove VLAs from the kernel. Google pays for a lot of nonsense...

> Even in c++ could you not just define some int [1000][1000]foo?

If it fits on the stack, yes.

Typical code using MD-arrays is scientific code, and the data they manipulate generally do not fit there.

Would the compiler not allocate the memory contiguously on the heap in that case then? Seems like a reasonable thing to do.

Nope. The C++ memory models is designed around no hidden/non-deterministic memory allocation.

If you try to allocate 10MB on the stack, that's the dev problem if the program fails, it's not the compiler job to guesstimate whether something will fit there or not (and it's impossible anyway, the compiler can't know all the stack sizes a program will ever run on).

I see. That makes sense.

I liked Objective-C (except the C parts). Such a breath of fresh air coming from C++ which was grown like a cancer with tons of features and you felt trapped by every one of them.

Objective-C in contrast was a very few additions thoughtfully added that composed cleanly and freed the programmer to actually get things done.

Scheme is (or at least was) coherent. You don't need to look any further than set/setf/setq to see that Common Lisp is "organically grown" from the fertilizer of a committee. CL does its best to make every other lisp more attractive.

Which Scheme are we talking about? R5RS? R7RS-small? R6RS? With SRFIs? Without? Which scheme? Is it `(library...)` or `(define-module...)`?

Heh, I'd probably take R4RS with define-syntax :-)

I mean, good choice, but you see the point, right? As much as ANSI CL has it's flaws, it has a standard, as much of a mixed bag it might be. Scheme is just a general potpourri of "we kinda have a guideline, but do whatever".

I would very much prefer scheme if the different implementations had a working standard. But I can't take my Chez-scheme code and throw it into Guile-scheme.

But pretty good chance I can take my ECL code and throw it into SBCL or LispWorks.

> you see the point, right?

Bah, I think this debate was already old when I first saw people arguing it on comp.lang.lisp in the 90s. I don't have a dog in this fight other than to reject the notion that Common Lisp is "coherent" and not "organically grown".

The original Scheme belongs in the category of languages like Standard ML and SmallTalk, where a small, careful, and talented group designed them with focus. Common Lisp seems like a bunch of smart people with competing interest and legacy baselines tried to meet in the middle. To the extent CL is more pragmatic, it's another example of "Worse is Better".

Scheme has a coherent and minimalist design, but its ecosystem and abstraction facilities feel too sparse for large applications.

When I started building a Lisp-based machine learning framework, Guile seemed like the right choice because it provides GOOPS and generic functions, yet I still ended up with a lot of boilerplate to compensate for the lack of a strong type system.

Scheme feels to me like C is to C++: not ergonomic for large-scale application development. Go is one of those languages that has both minimalism and productivity.

Common Lisp didn't grow into its warts, though, it was intentionally a "Common" Lisp. A compromise, like I wrote.

It isn't realistic to expect a design to be "proper in first place" because requirements change; my opinion is indeed the opposite - I find it natural for programming languages to have a (sort of) lifespan, and for new ones to (sort of) take their place.

Sure but literally everyone and their mom said these features were needed and then Go team said "nuh uh!!!" But, as it turns out, they are needed because they solve real problems, and are not just fake complexity like some people strawman.

Hopefully next they can add some error handling syntax and controls.

Indeed, in 2012, it was not clear to anyone that generics were needed /s

"Any sufficiently complicated C or Fortran program contains an ad hoc, informally-specified, bug-ridden, slow implementation of half of Common Lisp."

-- Greenspun's tenth rule

He had some lack of conviction to scope it so narrowly.

So which language had it right from the start? is there a language that has a very low rewrite status?

I think Elixir is a good candidate here. It's small, coherent, and composes well, and (at least to my understanding) the authors consider the language finished, with no new major features planned.

Elixir is missing static types though, it's hard to go back to work on dynamic languages.

Hopefully, it's coming.

https://elixir-lang.org/blog/2023/06/22/type-system-updates-...

I'd particularly like examples of statically typed languages that "got it right" (since I love me my types)

Ocaml maybe? Multi threading didn't seem necessary and introduced the possibility of data races.

That’s whataboutism - no language is perfect, but given when go released it’s fair to hold them to a higher standard than languages what were designed 25 years earlier.

As an aside - D, Zig, Rust, even typescript got most of the lessons learned from C right

I'm not familiar with D, but Zig and Rust are well-known for continuously evolving.

Zig has the (in)famous "Writergate": https://github.com/ziglang/zig/pull/24329

And besides Rust's high count of RFCs, there are things like async (I'm not complaining about it, but its an obvious large-scale "change"), module system changes, etc.

(To be clear, I like both languages a lot. But I wouldn't call them slow moving or right from the start.)

D literally can't even maintain backwards compatibility between minor version updates not to mention a big part of the D community left when D reinvented itself with D2. Among languages it's probably the one that is constantly in a state of flux.

what's a big project built with D? I feel it gets mentioned a lot on hackernews but I've never run into any project using it in the wild.

Flutter

Really? Wow, this is like Zig and I love Zig!

"properly designed" - ah yes, programming languages are famous for universally agreed upon design philosophies.

You may be tired of languages evolving over time, but there is no other way to build a rich and useful language.

ANSI CL is such a breath of fresh air nowadays. Does what you need, doesn't get in your way, comes with batteries included. And conditions are just god-tier.

> comes with batteries included

I like CL, but I can't agree that a stdlib that doesn't even have a string split function is batteries-included.

And to bring it full-circle, this is the exact same thing I run into with Go. When I mention how nice it is that Lang X has feature Y, someone is quick to point out that either, "You can BUILD that in Go" or, "You don't really need feature Y". We've proven that we don't really NEED compilers either... but I would hate to have to do my job without them.

so make your own and let's see how you do

I am actually working on my own language, and getting something better than Go is actually not that difficult!

The hard part about making a language is creating the stdlib and tooling and support for the language, but actually creating a language itself that has more features and better features than go can be done by a single person in a few months or a year probably, depending on how much experience they have.

Generics specifically are a great example here. A single person can implement a language with go-level generics fairly easily.

Have you?

There’s a fine line between being willing to change your mind and getting the basics wrong. Go has repeatedly gotten the basics wrong.

Declaring a highly successful language as having the basics wrong means that you are not correct about the basics that were needed.

Something can be highly succesful in spite of having glaring design flaws. Nobody is claiming go isn't wildly succesful, but it's _in spite_ of these issues. It was clear over a decade ago that iota, gopath, and lack of generics were massive kneecaps to the language; go changing it's mind on those things isn't progress it's just getting the fundamentals wrong.

A good example of where they're kind of stuck is date formatting - it's stupid, unclear, and likely a mistake, but it's not a fundamental flaw; it's just a quirk.

Why is iota a massive kneecap to the language? It is semantically identical to enum in C and Typescript.

The trouble is that Rust is older than Go and it was already confusing people into thinking enums and sum types are the same thing, so by using slightly different syntax, iota, Go avoided the whole confusion of users thinking that enums would behave like sum types instead of actual enums.

Is your attempt at making a point that not having sum types is the massive flaw? Sum types are a useful construct, to be sure, but there are plenty of good languages without them. That's more on the design quirk end, realistically.

> Why is iota a massive kneecap to the language? It is semantically identical to enum in C and Typescript.

iota is a massive kneecap _because_ it's semantically identical to enum in C and Typescript.

> Is your argument actually that not having sum types is the massive flaw? Sum types are a useful construct, to be sure, but there are plenty of good languages without them. That's more on the design quirk end, realistically.

In a dream world sure we'd have full blown sum types (and that would give a result type which would also solve a lot of the nil-interface-combined-with-error-handling issues that I've ran into when working with go), but I can forgive that. The problem is this - https://www.zarl.dev/posts/enums

> The problem is this - https://www.zarl.dev/posts/enums

The only case I see made in there is that it doesn't like how Go implicitly converts consts. While that may be a reasonable criticism, it doesn't have anything to do with iota. It is related to the type system and applies in general. Consider the same problem exhibited here:

    type Email string
    func Send(email Email)
    func() { Send("invalid") } // Converted string const does not satisfy Email type expectations

Perhaps you accidentally offered the wrong link?

It was made abundantly clear when Go was released that it was intended to "feel like a dynamically-typed language". Being able to pass arbitrary values is perfectly in line with a dynamically-typed language. Realistically, the type system in Go is there to give the compiler optimization hints, not to offer type safety. Go was targeted at those wanting to use Python, without the programs being painfully slow to run. How much of a kneecap is implicit type conversion, really, when it is already in line with what the target audience is accustomed to? It is a quirk at best.

> It was made abundantly clear when Go was released that it was intended to "feel like a dynamically-typed language".

If I google this quote a comment from you comes up here on this exact topic, where you seem to have completely missed the point there too. If I link to the docs [0], the full quote is "It's a fast, statically typed, compiled language that feels like a dynamically typed, interpreted language. " So it is a statically typed language first and foremost. If you want to rehash the discussion and tell people that a flawed type system that people have been asking for a solution to for close to a decade [1] you can just re-read the last time the arugments were made as I don't think I'm going to make any headway there.

[0] https://go.dev/doc/ [1] https://github.com/golang/go/issues/19814

> So it is a statically typed language first and foremost.

Right, because that is primarily how it makes things fast. Python is slow largely because it spends an inordinate amount of time trying to figure out what things are. Go knows what things are at compile time because the static type system tells it what things are and thus doesn't have to waste runtime compute on figuring out what things are, aside from when you use the reflect package, like Python does. That was its value-add — that it is kind of like Python, but faster. We already went over this...

> If I link to the docs [0]

I said original announcement, so I'm not sure why you wouldn't look there? Trying to be obtuse on purpose? Regardless, performance was indeed considered more important than being dynamically-typed. After all, if performance wasn't a concern then you'd just use Python. Go exists only because it was solving a problem that wasn't already solved. Slow Python was already solved. Type safety was already solved. It didn't need to go into those territories.

> Why is iota a massive kneecap to the language? It is semantically identical to enum in C and Typescript.

So is nil. Care to make the same argument?

Typescript has something resembling nil as inherited from Javascript, so presumably that's what you are referring to. C and Javascript are the most widely used languages in existence. Clearly they weren't kneecapped. It mightn't be to your preference, but "I don't like it" would be an atypical use for the word "kneecap".

There is a stronger case to be made for the other two. Calling GOPATH a design mistake is a stretch as it was perfectly suited to use within Google, but it didn't fit the typical solo developer's environment. Lack of generics made writing certain types of code difficult. You could be convincing in suggesting that Go did end up being used less than it otherwise would have because of those choices.

iota? It's just a construct that generates numbers (an enum). How does that kneecap anything? If it really bothers you, you can manually number the values by hand instead. Why would anyone reject a language because it allows you to optionally choose to have the compiler assign numbers automatically instead of forcing you to do it manually? The answer is nobody. In fact, most popular languages have something equivalent to iota.

Nil exists in C and is widely considered to be a massive mistake in the language. An understandable one at the time, but a mistake nonetheless.

Thus “go borrowed it from C, therefore it can’t have been a mistake” is a pretty lame take. The whole point of a new language is to make improvements on what’s out there already. Go missed an opportunity to fix one of C’s most notorious mistakes. So yes, they kneecapped themselves by forcing all of the users of Go to continue dealing with this well-known footgun.

Does it mean Go isn’t popular? Of course not. C was popular. PHP was popular. JavaScript is popular. Go is popular. This is always in spite of their faults. But Go could have been better.

> Nil exists in C and is widely considered to be a massive mistake in the language.

You're almost there, but it is wildly considered to be massive mistake in context of arrays. C has weird array semantics that are confusing and hard to get right, even for seasoned developers. That is where NULL comes to bite people time and time again. Go did not inherit C's arrays. Neither did Javascript. They go out of their way to avoid what C did. In Go, you can come close to the same semantics if you use the unsafe package, but take note the name.

Yes, they still have nil, but the scope is tightly constrained and while it is technically possible to misuse, you have to try pretty hard to do so. There are many other things that are more likely to end up being misused. Those would be the more massive mistakes.

> But Go could have been better.

Obviously. Every language ever created can be better.

> You're almost there, but it is wildly considered to be massive mistake in context of arrays.

From Tony Hoare: "I call it my billion-dollar mistake. It was the invention of the null reference in 1965. At that time, I was designing the first comprehensive type system for references in an object-oriented language (ALGOL W). My goal was to ensure that all use of references should be absolutely safe, with checking performed automatically by the compiler. But I couldn't resist the temptation to put in a null reference, simply because it was so easy to implement. This has led to innumerable errors, vulnerabilities, and system crashes, which have probably caused a billion dollars of pain and damage in the last forty years."

Odd that he wouldn't mention the word "array" anywhere in that quote.

> Yes, they still have nil, but the scope is tightly constrained and while it is technically possible to misuse, you have to try pretty hard to do so.

Like by not remembering to check if a pointer is nil? Or actually comparing one typed nil to a different typed nil?

> Obviously. Every language ever created can be better.

Only one of us is arguing that reimplementing C’s mistakes is actually a good thing.

> Like by not remembering to check if a pointer is nil?

Rule of thumb is to not introduce values that aren't valid. Equally, don't put in -1 for an age value, even if the language allows you to. You might later forget to validate that the age value is valid.

Yes, mistakes are possible, but these types of mistakes are pretty rare. There is some value in eliminating the possibility of those mistakes, sure, but we cannot pretend that it comes cost-free. There is good reason why almost nobody uses Rocq.

> Or actually comparing one typed nil to a different typed nil?

This is something that is likely to confuse, but not a facet of nil. It is related to interfaces. Let's not flail around like one of those wacky blow up things at the used auto lot.

Ah the good old “just be a perfect programmer and you won’t run into any issues” argument. Haven’t heard that one in awhile.

If only you had made it to the second paragraph.

Hey, at least we can now understand why you have such a hard time with nil, so that's something.

Your second paragraph doesn't hold after one minute spent reading the CVEs list.

Nil references in Go aren't likely to be a source of CVEs. Dereferencing a nil reference is not undefined behavior in Go; it's guaranteed to panic.

Not all CVEs are RCE. This one[1] for instance is go's nil-dereference caused.

[1] https://access.redhat.com/security/cve/cve-2025-29785

This is a bug that could also happen in Rust if someone used 'unwrap', or if the code panicked for any other reason. You can have logic bugs in any language. Remember the Cloudflare outage of Nov 2025.

It's also worth noting that the vulnerability you link to is a denial of service vulnerability. While yes, this is technically a vulnerability, it's not the sort of thing that people are usually worrying about in the context of null pointers. If features that give rise to DoS vulnerabilities are the ones to worry about, top of the list would have to be backtracking regex engines: https://owasp.org/www-community/attacks/Regular_expression_D...

Nil is not the problem, practically every language has such a concept - it's pretty much required to make things like Maybe/Optional work.

The problem is the implicitness of it and how easy it is to produce them and forget to check for them.

Rust is technically older than Go, but who was actually using it when Go 1.0 came out in 2012? Rust 1.0 wasn’t until 2015.

The social landscape doesn't depend on anyone actually using it. However, 1.0 isn't a significant milestone like you suggest either. For a current example, Zig is relatively popular today despite not yet reaching 1.0.

Do you just forget the things you write in earlier comments?

> Rust is older than Go and it was already confusing people into thinking enums and sum types are the same thing

Of course the social landscape depends on people actually using it. None of the people who weren’t using Rust at the time were magically confused about enums and sum types by the mere existence of some new and experimental language.

Rust barely existed at the time Go was first being developed. And given the history of Go and the notoriety of its core team for flatly ignoring prior work in programming languages, it’s extremely unlikely that Pike et al gave more than a cursory glance to what nascent Rust was doing at the time.

But even if they had, to suggest that they intentionally replicated a dumb thing from C but gave it a different name to avoid users being confused by a different thing from a language that roughly nobody knew about at the time is bananas.

> Of course the social landscape depends on people actually using it.

That's nonsense. Brainfuck has shaped the social landscape despite effectively nobody using it, and absolutely nobody using it for any real work. The social landscape is not at all dependent on use.

> And given the history of Go and the notoriety of its core team for flatly ignoring prior work in programming languages

Huh? Go comes straight out of prior work. It is nearly indistinguishable from Alef. What the Go language flatly ignored was being innovative. Reasonably so, of course. It wasn't trying to innovate in programming languages so that we'd have another to throw on the heap of languages nobody uses. It was trying to solve a specific business problem using well-established methods.

It's a highly successful language because (1) it was backed by Google, and (2) created by Robert Griesemer, Rob Pike, and Ken Thompson.

If it came out of anywhere else, it might have struggled even to hit the homepage here.

This logic is easily shown to not hold. Why isn't Carbon, Dart, etc. not really popular then?

I can't speak about Dart, but Carbon had just barely started development when it was first announced 4 years ago, and is currently presented as an experimental language that is not yet ready for use [0].

0: https://github.com/carbon-language/carbon-lang#project-statu...

Dart was relegated to effectively being Flutter. It was originally intended to supersede JavaScript but no other engine would commit to that. It failed at its initial goal and hasn’t really found a footing outside of Flutter.

Carbon is by its own admittance not ready to use and I think mostly relegated to solving Google’s problems with C++ right now.

Both of them didn’t ship with a standard library as robust as Go’s.

One thing that made Go popular out of the gate is it is extremely good fast to build out robust HTTP services and infrastructure.

This is a very common use case and they tailored Go to be a great fit for it. You can build your entire backend without a single third party module if desired using Go’s standard library and it isn’t terribly complicated to do so.

Its just bitter dorks bitter their pet language with cutting edge programming abstractions didnt make it to the big leagues.

By that logic Windows would be the best operating system ever and perfect in every way, and anyone who disagrees must be wrong about how an OS should be.

And Javascript and Python the best languages.

The basics of a programming language were wrong. The basics of marketing were very right. Those are not the same.

So you mean to say that PHP5 and Js from 2007 had a well-founded design?

https://en.wikipedia.org/wiki/Worse_is_better

Yes, all great examples of language design. PHP, Java, JavaScript, C, C++, Go

An engineer, of course, understands that there is no such thing as "wrong", only different tradeoffs, but with the rise of "vibe coding" you don't need to be an engineer to play in the world of programming anymore.

cough JavaScript cough

Sounds like you want this feature, and you just got it. Not sure how that's wrong. You don't add in every feature from the start.

I wanted it 10 years ago.

I don't think anyone admitted any wrong or had any big change in philosophy. It's always a good thing to learn something along the way. But the current message seems to be that this was the plan all along, and it just took some time to design properly.

Of course adding generics is not something that every language needs to do. Scripting languages like Ruby don't really need this style of generics. It doesn't fit the design of the language, and it's not even clear what that would look like in Ruby.

But static typing with generics does solve a recurring problem, and we've seen some real convergence towards type hints and type systems even in staunchly dynamic scripting languages. Modern Javascript is now mostly Typescript, and they've successfully retrofitted a very advanced type system in the last place I would have expected 20 years ago.

Type hinting seems like the worst of both. You pay the cost on refactor to go change them all, where dynamic typing or static type inference avoid that. You also don't have any of the benefits of static or dynamic typing. My strong preference is static typing with good inference and an ide that shows the inferred types everywhere when asked. Dynamic typing can make some tasks dramatically easier, I'm just not capable of using them without making hideous mistakes.

It’s still annoying ~20 years after Java did the same mistake of not including generics, which was already clear to many people with C++ experience back then.

...and Java didn't even have basic enums or sum types from the beginning. But it had null.

They added enums, they added sealed classes. They're trying to get rid of null (apparently it's really hard). The problem is that in 2012, when go 1.0 was released, this should have been obvious to everyone.

Here's a famous discussion from 2009, three years before the 1.0 release (tldr: facepalm)

https://groups.google.com/g/golang-nuts/c/rvGTZSFU8sY

I remember back in 1995 thinking that it was stupid for Java not to have generics, so instead you had to always cast Vector/Hashtable elements from Object, or implement your own type-specific container classes for every element type (and there wasn’t even a preprocessor to facilitate the latter).

Sum types I didn’t really miss, because you can implement a type-safe equivalent using the Visitor pattern, and retain an interface-implementation separation that native sum types typically don’t provide.

That is an interesting read, seems some were having a hard time grasping the benefit of having compiler checks for potential null dereferences. Having worked with null safety in TypeScript and Kotlin the extra bit of strictness is nice.

The biggest design issue with adding value types and non nullable to Java, is that the number one design requirement of any solution is not to break Maven Central.

Every compiled JAR out there has to keep working as always on a JVM with updated semantics, and worse code has to be compatible, when passing class instances around between old and new code.

Then there are the guest languages on the JVM as well.

Worst thing a programming language can do is introduce core semantics changes after 1.0. See Python 2 -> 3 and Zig's *gates.

Except Zig is not 1.0

Of course, just a good case study of how radical languge changes are received by users.

Because they refuse to do something until they figure out how to do it well.

I respect that.

Go's generics design is the most clunky one among popular languages.

Exactly. We heard for years they wouldn't do generics it until they could do it right, and that was perfectly fine with me. Who wouldn't want a well thought out implementation? Then they released generics and it was like, "this is what you thought was the right way to do it?!"

What’s the approach they should have used instead and how would it be better? Especially in terms of keeping fast compile times and overall performance.

Of course, if you go back and watch the original Go announcement it said that it would need generics once they figured out how to do it. And when the first version of generics landed it was said that generic methods would be added later, once they figured out how to do it. So that isn't applicable here. The need was always recognized.

If Go had just taken an off-the-shelf implementation of generics in 2009 then they could have spent the last 16 years deliberating over something useful, rather than attempting to reinvent programming language theory.

The impression I have always gotten from Go's designers is that they are rather arrogant and averse to the idea of using other people's work. They want to develop everything from first principles, but by so doing end up with poor reinventions of well-studied concepts.

> and averse to the idea of using other people's work.

They did use someone else's work, though. If you recall, Philip Wadler (of Haskell fame) designed Go's generics.

> but by so doing end up with poor reinventions of well-studied concepts.

Which is funny as there is probably nobody on earth that would be more capable than Wadler to get the job done. His pedigree in that area of work is pretty astounding. If he couldn't do more than create a poor reinvention, what hope did the laymen working on the Go core team have?

Answer: They had no hope. It's not like they weren't trying. Ian Lance Taylor, for instance, is well known for beginning work on generics in Go before it was even first released to the public. He, among others, quite simply, were unable to figure it out.

Everything looks easy and straightforward when observed comfortably from an armchair, I suppose.

Yes, the same guy that supported the effort to add generics to Java, a decade beforee Go came to be, talk about not getting language design history.

Stop excusing them, they were the first to acknolowdge being wrong in first place,

-- https://go.googlesource.com/proposal/+/master/design/go2draf...

> Stop excusing them

What is there to excuse? Your quote confirms that they simply don't know what they're doing as was already established. Not that anyone should expect them to. They're just regular average humans, same as every other random Joe you encounter while walking down the street, who all equally have their own failings and shortcomings. Why HN is constantly trying to put these particular people on a pedestal, I'll never know. Jealously that regular bumbling idiots just like them accidentally stumbled into creating something popular (for some definition of popular), perhaps?

They certainly knew what they were doing.

The first step to acknowledge not being great language designers is to actually be humble to learn from the work of others.

Not assert to all "we know better", and then repeatly having to backpedals on such assertions.

> They certainly knew what they were doing.

Based on what? Sure, the one guy managed to hack up the first version of C, but it took a complete rewrite and some major changes to become what we would actually recognize as C. Not exactly indicative of someone who knows what he is doing. And everything in Go was copied straight from Winterbottom's work.

Where did "they" say "we" didn't need generics? That sounds like a bad faith / misinterpretation / straw man; as someone else pointed out, they postponed generics until they figured out the use cases and whatnot.

Remember that the generics implementations in other languages (like Java) take up half the spec + implementation - that's not something that Go wanted.

From another commenter here:

> The post quotes the Go FAQ as saying, "we do not anticipate that Go will ever add generic methods".

You keep posting that. Do you understand the difference between that and "we anticipate that Go will never add generic methods"? What they actually said shows epistemic humility and recognizes that they might change their mind.

Hi! I'm a human being that is trying to understand the world and make friends along the way. I see you are too! Pleased to meet you.

You asked the question

> Where did "they" say "we" didn't need generics?

And I (re)posted a quote from them, which sounds to me like, at the time, they believed that "we" Go users didn't need generics.

They may have changed their mind, which is totally fine! But I do think it sounds like the person you were replying to wasn't commenting in bad faith or misunderstanding or fighting a straw man as you posted. Seems like a reasonable interpretation of what the Go devs had said at one point. To each his own though!

Hi! If you want people to think you are a human, you should not mistake one of them for an entirely different one, and you should not refer to yourself in the third person.

Ah, I guess to err isn't human after all :)

Do you understand the difference between generics and generic methods?

We already had generics when they wrote "we don't anticipate adding generic methods."

The OP didn't mention generics or generic methods :) They referred to something that was added, after the Go team didn't anticipate adding it. Sounds like a fair point to me!

What did they say about proper enums.

You already have iota. Type safety is not needed by design:

> Go intentionally has a weak type system... Go in general encourages programming by writing code rather than programming by writing types...

https://github.com/golang/go/issues/29649#issuecomment-45482...

Theres a whole universe between proper enums and go becoming Ada. Same with null safety.

Ada is the lost knowledge of a past, more advanced civilization.

Thankfully not everyone has forgotten.

https://www.adacore.com/case-studies/nvidia-adoption-of-spar...

complaining about things given to you for free

I frankly don't buy into this trope that a lack of monetary cost should shield something from criticism. Anything created by humans for other humans, especially tools meant for getting work done, should certainly be open to evaluation/judgement/critcism, regardless of whether the creator chooses to charge for it.

And it's not like Golang is some freshman student's hobby project; it was created by one of the world's largest tech companies, by people with a strong pedigree in programming language design.

I remember lack of generics being pitched as a feature of Go initially, not a lack. The original design goal was simplicity. I don’t use Go, so have no opinion on this, just interesting that it’s going in this direction.

Generics being omitted wasn't pitched as a feature that I ever saw. The discussion was mostly around "hic sunt dracones" with generics. The Go authors were very aware from the get-go that generics are an incredibly difficult thing to implement well, and the commitment was always to not do it until they were really ready to tackle it. The bizarre thing to me is that when they did tackle it, they left this massive implementation gap, and then what little they did implement didn't seem useful enough that anything meaningful was added to the standard library. So all this talk about generics for years was a big nothing-burger. That is, I hope, until 1.27 when this gets released. Flags is one in particular I think is in desperate need of proper generic methods.

This is great. Will be useful for data access methods!

As for the detractors, from the first generics proposal this was called out as a "not now", not never. There were questions of implementation. They aren't a super large team, and they try to do things incrementally and do them well.

> As for the detractors, from the first generics proposal this was called out as a "not now", not never.

What? The post quotes the Go FAQ as saying, "we do not anticipate that Go will ever add generic methods". There is also some similar discussion of the original generics proposal, with language like "then it's much less clear why we need methods at all". (I'm omitting some context, but I don't feel that it changes the meaning.) Those feel much closer to "never" than "not now".)

The post is also subtitled "A change of view".

No, you’re clearly wrong; golang was always going to add support for generic functions.

Everyone also wanted and accepted the need for generics. It was always something they wanted to add to the language. Rob Pike never said that that kind of abstraction isn’t what golang is for. It was always just a matter of getting the design right.

Go has always been a systems language. It was one when we thought it was going to fit nicely for low-level, high performance use-cases. Given that the GC, runtime overhead, lack of control over memory layout, and other issues made it a poor fit for what we historically thought were systems language tasks, it’s still a systems language because we’ve grown to understand that the term “systems program” has always meant network middleware that shuttles around JSON and transforms it.

Dependency management too. Modules were something that nobody argued were unnecessary. None of the language developers ever claimed that “you should always build against HEAD, and if upstream breaks you, that’s a coordination problem to be solved socially”. The community didn’t need to independently invent godep, then glide, then govendor, then dep, before the core team finally shipped modules. That was just enthusiastic parallel exploration of a problem space that everyone agreed was a problem.

GOPATH was always understood to be an awkward temporary scaffold that everyone tolerated while the real solution was being designed. The single-workspace model was never defended as philosophically correct or a deliberate feature of the language. When modules arrived, everyone was simply relieved that this obvious stopgap was finally replaced.

The core team always intended to add builtins for min/max. Nobody ever told you to just write `if a > b { return a }; return b` yourself because it was “only two lines.” The fact that every Go codebase in existence had its own copy of this logic, typically buried in a file called util.go, was not evidence of anything being missing from the language.

Range was always a stopgap before iterators could be implemented. Nobody ever argued that iterators were needlessly complicated and went against the spirit of the language. The slices and maps packages provided important missing features that everyone using the language wanted.

Everyone agrees that errors were anemic from the outset. errors.Is/errors.As are nice additions but everything was Just Fine™ before they were added.

Speaking of errors, having two lines of error-handling boilerplate for every line of code is good, and right, and perfect. It’s not verbose; it’s “explicit”. But when that gets changed to be less verbose, we will all agree that it was always a pain and made reading code unnecessarily more difficult and that everyone always expected this to be fixed some day.

I personally can’t wait to see what next development will never have been “against the Go philosophy” and definitely not something that gophers argued was perfect the way it was any time misguided malcontents and rabble-rousers wrongly tried to suggest the language wasn’t perfect the way it was.

> The community didn’t need to independently invent godep, then glide, then govendor, then dep, before the core team finally shipped modules. That was just enthusiastic parallel exploration of a problem space that everyone agreed was a problem. When modules arrived, everyone was simply relieved that this obvious stopgap was finally replaced.

And of course, it was replaced with a more correct implementation that was incompatible with that awful stopgap because semantic correctness trumps all. vendor/ trees and GOPATH were never meant to be remotely compatible, and don't you know -- the Go compatibility guarantee(TM) doesn't apply to misuse of GOPATH to work around shortcomings^Wwell-considered designs of Go, even if it breaks the largest Go project at the time!

(/s It still shocks me that they decided to drop "src" from vendor/src and break compatibility when they finally got around to supporting vendoring despite every tool using it. And symlinks don't work because Plan 9 is the future!!)

> term “systems program” has always meant network middleware that shuttles around JSON and transforms it.

Who are we that has always defined that term that way. For any systems programmer golang has pretty much not been a solution.

Systems is below layer 4 of the network stack, it is building the network stack in the first place.

I think your sarcasm detector has gone missing chief.

"Systems program" or "systems programming" is a meaningless term because people use it differently. Discussing what it means is a waste of time. It is better to be specific.

Um, there is user-level implementation of TCP/IPv4+IPv6 stack in Go [0], developed and used by Google as part of its gVisor.

[0] https://github.com/google/netstack

I love your rant.

The writing is on the wall for next development.

“For the foreseeable future, the Go team will stop pursuing syntactic language changes for error handling. We will also close all open and incoming proposals that concern themselves primarily with the syntax of error handling, without further investigation.”

You have those who obsess over feature lists for languages and then you have those who obsess over friction, productivity, ease of use and developer ergonomics.

If I were uncharitable I might call the categories "people who are somewhat removed from reality" and "people who inhabit observable reality". Avoid the former, treasure the latter.

Languages evolve for a reason and nobody should give a shit about people who do not understand why.

> I personally can’t wait to see what next development will never have been “against the Go philosophy” and definitely not something that gophers argued was perfect the way it was any time misguided malcontents and rabble-rousers wrongly tried to suggest the language wasn’t perfect the way it was.

I really hope it is more ergonomic error handling. Or maybe sum types/discriminated unions.

Those are fine the way they are. I don’t want the language getting more complicated. Rob Pike has said they’ll never happen. Suggestions like this are extremely against the Go way of doing things. Actually the verbosity is a good thing. It’s explicit.

> I really hope it is more ergonomic error handling.

They've publicly said no more language changes specifically for better error handling are coming, it will at most be the library-level improvements.

So this would be in line, then, with all of the other features they’ve publicly said aren’t coming and aren’t necessary but then do anyway after retconning their original stance?

Could esteemed commenters please read the words the Go language team actually wrote?

    > Or, we could decide that parameterized methods do not, in fact, implement
      interfaces, but then it's much less clear why we need methods at all. If
      we disregard interfaces, any parameterized method can be implemented as a
      parameterized function.
    
    > So while parameterized methods seem clearly useful at first glance, we
      would have to decide what they mean and how to implement that.

Well, they have finally decided what parameterized methods actually mean, and they see how to implement that, and it all seems clearly useful. So...

> No, you’re clearly wrong; golang was always going to add support for generic functions.

Let's get this straight. I'll give you a long quote from Rob Pike's article where he describes the history of the go language:

""" One thing that is conspicuously absent is of course a type hierarchy. Allow me to be rude about that for a minute.

Early in the rollout of Go I was told by someone that he could not imagine working in a language without generic types. As I have reported elsewhere, I found that an odd remark.

To be fair he was probably saying in his own way that he really liked what the STL does for him in C++. For the purpose of argument, though, let's take his claim at face value.

What it says is that he finds writing containers like lists of ints and maps of strings an unbearable burden. I find that an odd claim. I spend very little of my programming time struggling with those issues, even in languages without generic types.

But more important, what it says is that types are the way to lift that burden. Types. Not polymorphic functions or language primitives or helpers of other kinds, but types.

That's the detail that sticks with me.

Programmers who come to Go from C++ and Java miss the idea of programming with types, particularly inheritance and subclassing and all that. Perhaps I'm a philistine about types but I've never found that model particularly expressive.

My late friend Alain Fournier once told me that he considered the lowest form of academic work to be taxonomy. And you know what? Type hierarchies are just taxonomy. You need to decide what piece goes in what box, every type's parent, whether A inherits from B or B from A. Is a sortable array an array that sorts or a sorter represented by an array? If you believe that types address all design issues you must make that decision.

I believe that's a preposterous way to think about programming. What matters isn't the ancestor relations between things but what they can do for you.

That, of course, is where interfaces come into Go. But they're part of a bigger picture, the true Go philosophy. """

Rob Pike, 2012

I can draw a few conclusions from this: firstly, he didn't want to add generics at all because he didn't think they were useful, and secondly, he doesn't understand programming very well and doesn't know what generics are and confuses them with inheritance.

I could be mis-remembering it. I didn't look up and src it.

Gophers are usually quite fast, perhaps an elderly turtle would be a better mascot?

In day-to-day usage, the (fast) compilation speed matters much more than the (slow) implementation of new features.

I totally agree, but I'd go further and argue that slow implementation of new features is itself a desirable trait. It's one of the reasons why why I like both Go and Clojure.

Spot on. Heaven forbid it turns into a C++ (I'm not a Rust practitioner but from the outside it seems to accrete features pretty quickly as well).

The ease of grokking Go (both reading and writing) are big advantages, and facilitated by the "small" feature set of Go.

> Gophers are usually quite fast, perhaps an elderly turtle would be a better mascot?

The slow turtle wins the race against the overly eager rabbit... so I'm okay with that

The lesson of the parable of the tortoise and the hare isn't that "slow beats eager", it's a lesson against unfounded overconfidence in one's own abilities.

Chasing a perceived gap between language features and user expectations has been and continues to be the greatest error in the leadership of Go.

The nagging imperative requires a stronger response than the capitulation of identity.

They should have blocked generics forever and instead have a proper enum. Unpopular opinion, I know.

Since they can't implement interfaces, Generic methods are just syntax sugar for generic functions. I'm surprised they actually accepted this proposal for sugar.

Yes, the sugar is just to make chain calls with parameter types possible. The sugar reflects the limitation of the basic of Go generics design. Now they would make the language even more complex for such a small need. In facts, there are more problems in Go generics need to be solved earlier than this: https://go101.org/generics/888-the-status-quo-of-go-custom-g...

Thanks for sharing. I had no idea Go's generics had this many limitations.

This resolves a big gap in generics for most people coming from other languages to go so I completely approve this direction. Not saying use it everywhere but if you must use it, it’s better to have it on the struct than call a module level generic func.

Watch as this gets rejected for nonsensical reasons just like the issue about adding uint128, which also has been open for an extremely long time and they keep moving the goalposts as to reasons why it can't be done...

It’s accepted.

Now there's a surprise. I've generally been very disillusioned with Go after they absolutely stonewalled everybody on uint128 (and continue to do so) for absolutely no reason (and ignoring that it would make many things in the language easier to express).

> it would make many things in the language easier to express

Like what?

Like IPv6 addresses, UUIDs, the list goes on. Does it mean anything that Go themselves had to invent a custom uint128 type in the standard library because they didn't want to add it to the language? There's a very long list of instances of them stonewalling it here: https://github.com/golang/go/issues/9455

I agree with their reasoning, you don’t typically do math with a UUID, 128 bit cryptographic hashes are insecure and outdated anyway, and IPv6 is kinda valid but rare and still doable with slices.

I wrote Go professionally for years. I don't know how many hours I've spent debugging and reviewing nil pointer bugs and race conditions. Generics were sorely needed but the initial implementation was lacking.

I moved to Rust professionally 4 years ago and haven't looked back. Mutex<T> Option<T> Result<T, Err> are all phenomenal.

I've written everything from web backends, frontends (hurry up wasm, seriously), to Node.js and Python extensions.

Web backends use under 1mb of memory and can support hundreds of thousands of concurrent users on a $2/m VPS. Frontends can be beautifully multithreaded. Native extensions can dance between OS threads and multi-threaded runtimes.

When I review code I focus only on the logic, not sidetracked by reasoning about race conditions or anything. Great when you review the work of less experienced contributors.

The ultra strict compiler is extremely helpful with LLMs. You bounce back and forth until it compiles and, if it compiles, it's usually correct.

It's at the point where I can't really see a use case for another language - and yet, no one uses it! It's madness!

I have nothing to add to your comment, but it matches so eerily my exact experience, that I just wonder if your alias is a second account of mine :-)

By frontend, do you mean wasm in the browser? I'd have expected multithreading there, especially in wasm, to be extremely unergonomic.

Maybe you mean to refer to concurrency?

What a happy surprise today! The amount of times I’ve had to do weird janky package APIs so the API was still reasonable is more than I can count.

OMG. I'm going to recode some of my libraries.

And so the cycle will continue. Always a shame when languages cave like this and add extra unnecessary complexity and error prone hard to parse syntax.

It'll be interesting to see the next language that comes along rejecting bloat in favor of simplicity, and then we can all start again.

what's wrong with `#define`? <laughing hysterically>

nah I'm kidding

Seriously, what's wrong with `#define`?

A sad day for Go, the pHDs have won, simplicity has died.

Agreed 100%. https://itnext.io/go-evolves-in-the-wrong-direction-7dfda8a1...

It certainly makes readability of Go code harder.

If generics replace code generation or hand coding a concrete implementation for specific types or... operating against interface{} and then switching based on runtime type, then I strongly disagree that it makes reading Go code harder, it makes it, IMO, easier.

But that's the thing, it's just IMO.

It only died if you actually applied it in your own codebase - as with any feature, using it is optional.

Code is written to be read, and typically for other people to read. So, in practice, it is only optional if everyone else agrees to not use it.

Reply on news.ycombinator.com