I find this comment from Griesemer [0] on one of the github issues for enums in Golang quite insightful:
>> [...] all the proposals on enums I've seen so far, including this one, mix way too many things together in my mind. [...] Instead, I suggest that we try to address these (the enum) properties individually. If we had a mechanism in the language for immutable values (a big "if"), and a mechanism to concisely define new values (more on that below), than an "enum" is simply a mechanism to lump together a list of values of a given type such that the compiler can do compile-time validation.
Like with generics, I like the team's approach of taking features seriously, not adding them just because other languages have them, but actually trying to figure out a way for them to work in Go, as cleanly as possible. I think computer science, as a field, benefits from this approach.
And I also dislike many things from Go, and I want "enums" badly too, but that's for another comment.
[0] https://github.com/golang/go/issues/28987#issuecomment-49679...
When there’s no clear winner in terms of tradeoffs, I prefer to leave it out of the language like Go has done.
Yes, they are poison for the evolution of a public API.
I think that Go's decision stems from protocol buffers as they allow to push new values through old binaries, which is a must once you grow enough.
https://developers.google.com/protocol-buffers/docs/proto3#e...
When clients use `switch` on a non-frozen enum from outside its defining module, Swift emits a warning if they don't have an `@unknown default:` case... so consumers of your enum will have to have default logic for handling new cases in order to avoid this warning. (Not for frozen enums though, for frozen ones it's enough to just cover the known cases in calling code, since the expectation will be that you can't update them.)
So basically, if you don't bother thinking much about the problem, you can just avoid adding `frozen` and you'll probably get reasonable behavior where you can add more cases later. Using `frozen` should only be the case if there is some sort of logical impossibility for there to be more cases. Something like how `Optional` has .some and .none, but it's pretty obvious that nobody's going to go add a new case to it (what would a new case even mean?) Same with Result, and probably a bunch of other types I can't think of at the moment.
Also worth noting that Swift treats intra-library code very differently than code that links from another library... if you use your own enums in your own module and don't make them public, it treats them as if they're always frozen... which is nice because it's your internal code and you can always update your own usages without having to worry about compatibility.
Yeah, non-frozen by default makes a lot of sense. The only gotcha left is that you can't retract from adding frozen, but that's ultimately behavior you want and something that must be able to bite you back.
> if you use your own enums in your own module and don't make them public, it treats them as if they're always frozen... which is nice because it's your internal code and you can always update your own usages without having to worry about compatibility.
That's neat
Unless your services talk to each other or share some external data storage? Which is actually really common?
What I mean is, code that deals with serialization cannot by definition avoid the problem of “what if the data is invalid”. It’s not just enums but every aspect of your type system that must deal with this problem (typically by just throwing an error if the data is invalid, etc.)
I honestly think we’re talking about different things… the guarantees a programming language gives you are independent of the guarantees a serialization format gives you.
https://doc.rust-lang.org/reference/attributes/type_system.h...
I'm still not sure if it's worth it, it's idiomatic Go to "fall-through" if-branches for default cases, which is the same when checking quasi-enums. The symmetry is nice and makes it very easy to read. But I could be convinced.
In Rust, the answer is #[non_exhaustive], which forces consumers to always add a default case. It’s not a huge deal, just a known issue with a well-understood solution.
If something is handling A and B, but you add C, the code probably needs to make sure it's handling C correctly.
I use Java in my dayjob and the behavior you've outlined is how I always code things, but it's manual and doesn't happen at compile time: I provide default that throws a runtime exception.
I'm not disparaging closed enums, they are very useful in certain contexts, but they make it really easy to do the wrong thing when reading data off the wire. Given Go is focused on this exact domain (distributed systems), I am glad the language doesn't have them.
What you want here is something akin to Rust's match behaviour on enumerated types. If your alternatives aren't exhaustive, it doesn't compile. Now, Rust is doing that because match is an expression. Your day matching expression needs a value if this is a Thursday, so not handling Thursday in your day matching expression is nonsense - even though often the value of a match isn't used and might be the empty tuple it necessarily must *have a value.
It seems to me that today a Go Switch statement operating on days of the week can omit Thursday and compile just fine. Exhaustive switch means that's a compile error. If your "exhaustive switch" is optional or just emits a warning, it won't catch many of the problems for which exhaustive switch is the appropriate antidote.
This would only make sense on an enum type, which would be a completely new thing, so it can be introduced without breaking backward compatibility. Constants and switch on non-enum values would stay, because they are useful independent of enum types.
However, zero values throw a wrench into this. The zero value of an interface is nil, so enhancing interfaces would require you to address what happens with an uninitialized variable. One of the current proposals suggests that nil continue as the zero value.
They could introduce a totally different type, like a sealed interface, which doesn't require a zero value, but that distinguishes between different types of interfaces, and I'm not sure how that'll be received.
Beware, tho, that with many languages today you’re not really doing that even when they advertise enums e.g. in both C# and C++, enums are not type-safe (not even `enum class`). Iota is, at least, a fair acknowledgement of that.
> with a FromString() function
That seems like way a step too far, is there any such “default method” today? And I don’t think Go has any sort of return-type overloading does it?
Have you considered reading?
By allowing any value of the underlying type, even if they were not values of the enum. The language most famous for this is obviously C:
enum foo { A, B, C };
int main() {
enum foo x = A;
printf("%d\n", x);
enum foo y = 42;
printf("%d\n", y);
}
Now you might argue that this is C so obviously it fucks this up, however that C does this leads to further issues:
- For compatibility with C, C# enums have the same property. I don't know about you, but that surprised me a great deal since Java has type-safe enums (even if they're not sum types).
- Even more so, C++ added `enum class` in C++11, and obviously has inherited C's enums, but enum class still is not type safe, afaik the differences are that `enum class` is scoped (so in the snippet above you'd need `foo::A`) and it's not implicitly convertible with the underlying type. But it's still explicitly convertible (via a cast or list initialisation), meaning you can't assume a caller will remain within the enum.
Check the upvotes on my comment.
[1] https://en.wikibooks.org/wiki/Ada_Programming/Types/Enumerat...
> We use enums heavily to force devs who use our code into good choices
the note is that this is very much language-dependent and there are languages which get it very, very wrong.
> And I'm sure there are languages that do enums "properly", but it's always C/C++ that is referenced.
That makes no sense, the original comment obviously assumes a language which does it "properly".
type SignedInteger interface {
~int | ~int8 | ~int16 | ~int32 | ~int64
}
> We have proposed that constraints can embed some additional elements. With this proposal, any interface type that embeds anything other than an interface type can only be used as a constraint or as an embedded element in another constraint. A natural next step would be to permit using interface types that embed any type, or that embed these new elements, as an ordinary type, not just as a constraint.
> We are not proposing that today. But the rules for type sets and methods set above describe how they would behave. Any type that is an element of the type set could be assigned to such an interface type. A value of such an interface type would permit calling any member of the corresponding method set.
> This would permit a version of what other languages call sum types or union types. It would be a Go interface type to which only specific types could be assigned. Such an interface type could still take the value nil, of course, so it would not be quite the same as a typical sum type.
> In any case, this is something to consider in a future proposal, not this one.
This along with exhaustive type switches would bring Go something close to the sum types of Rust and Swift.
https://github.com/BurntSushi/go-sumtype is great, but a bit unwieldy. Language support would be much better.
func UnmarshalOneOf(data []byte, args []interface{}) (index int, err error)
Which I use like this:
variants := []interface{}{ &T1{}, &T2{}}
i, err := UnmarshalOneOf(data, variants)
// …
return variants[i]
This is what linters will do for you by default.
> we instead need to create ValidEnumValues and remember to update it every time a new enum value is added
Code generators are first class citizens in go, and writing an icky but reliable test won't be too hard either.
Create a new int type and use that for your enums. While you still can create an illegal enum value, you basically have to be looking for trouble. It’s not going to happen accidentally. It’s even harder if it’s an unpunished type in a different package.
See:
https://github.com/donatj/sqlread/blob/91b4f07370d12d697d18a...
type Test int
const (
T1 Test = 0
T2 = 1
)
func TestSomething(t Test) {}
...
TestSomething(17)
package main
import "fmt"
type Test int
const (
T1 Test = 0
T2 = 1
)
func main() {
t, err := TestString("T1")
if err != nil {
panic(err)
}
TestSomething(t)
}
func TestSomething(t Test) {
fmt.Println(t.String())
}
Like C or C++ do? :)
Other languages take a stricter approach, and maybe that's better. Not defending (although I like Go), but it's really more a language philosophy than a singular defect.
As the other commenter noted, this should fail code review and you should be using the provided constants, and it should be clear to you. And if you disagree (which again is totally valid), you should use a stricter language—there's plenty out there!
I see your point however.
Also:
variableDefinedInAFarAwayModule := 17
...
TestSomething(variableDefinedInAFarAwayModule)
variableDefinedInAFarAwayModule is definitionally type int and will not be cast. It is also unpublished, so you couldn't be using it for a faraway module?
Your 17 in the previous example has it's typed determined at compile time which is why it can be a problem.
TestSomething(T1 & T2)Int types also don't give you any guards when deserializing.
We use enums heavily to force devs who use our code into good choices, but the options are currently:
1) Use int-type enums with iota: no human-readable error values, no compile-time guard against illegal enum values, no exhaustive `switch`, and no autogenerated ValidEnumValues for validation at runtime (we instead need to create ValidEnumValues and remember to update it every time a new enum value is added).
2) Use string-type enums: human-readable error values, but same problems with no compile-time guards, exhaustive switch, or validation at runtime.
3) Use struct-based enums per https://threedots.tech/post/safer-enums-in-go/ : human-readable error values and an okay compile-time check (only the all-default-values struct or the values we define), but it still doesn't have exhaustive switch, is a complex pattern so most people don't know to use it, and suffers from the mutable `var` issues the post author detailed.
To my naive eye, it seems like a built-in, compile time-checked enum type with a FromString() function would help the community tremendously.