Is it natural for me to have skeptical feelings for something like this?… The reasoning being, there has already been so many man hours poured into FP languages, by various experts, and it hasn’t been “paradigm changing”. On top of that, optimizing FP seems to take a lot of work and knowledge. You’re telling me they’ve got all the experts, and hammered something out (which itself sounds inadequate to achieve something great here), which is better than GHC or similar? It’s incredible if true. Like insanely incredible. I would love for someone to explain to me why Verse has a chance at all.
… The reasoning being, there has already been so many man hours poured into FP languages, by various experts, and it hasn’t been “paradigm changing”.
It definitely it has been, look around! Languages have been copying lots of things from FP tools. Algebraic data types, higher-order functions everywhere, pattern matching, immutable by default, futures and promises, generators, etc. These features are many decades old in Haskell, OCaml, etc.
FP and these experts have already changed the paradigm.
You’re telling me they’ve got all the experts, and hammered something out (which itself sounds inadequate to achieve something great here), which is better than GHC or similar?
No, it’s completely different. Verse is absolutely definitely not Haskell and never will be. The goal of GHC is totally different to what Epic Games is doing with Verse.
People in Epic Games have been thinking about this for decades, recently invested a bunch of time, hired a bunch of experts, made a language, decided to rewrite one of the most popular games in the world using the language they just made. So yeah, I think “Verse has a chance” at doing what it was exactly designed to do. Why not?
decided to rewrite one of the most popular games in the world using the language they just made
Verse’s current implementation compiles to Unreal Engine’s Blueprint VM, so I imagine the integration into Fortnite was simple and not a complete rewrite. The /Fortnite.com/ package contains a bunch of Fornite specific APIs that FII out to Fortnite code iiuc
Yes, the initial integration uses existing tools, but from Tim Sweeney (just after 33 minute mark):
We see ourselves at the very beginning of a 2 or 3 year process of rewriting all of the game play logic of Fortnite Battle Royale (which is now C++) in Verse…
Verse’s current implementation compiles to Unreal Engine’s Blueprint VM
I really hope the Open™ implementation is implemented on top of something a lot more efficient. Blueprint is horribly wasteful. The VM and bytecode has been largely the same since the UnrealScript days - for the curious, it’s essentially a glorified tree-walk interpreter (but with compact bytecode instead of heap-allocated nodes.) It really makes me wonder why Epic decided to continue using the same inefficient tech for UE4’s Blueprint.
The guy who presented the language on stage built a lot of JavaScriptCore / WebKit internals, so they certainly have the right people to build a very fast VM.
Languages have been copying lots of things from FP tools.
I should’ve been clearer in what I meant, but this is definitely true! Definitely true… What I meant was any particular FP language hooking everyone into the FP paradigm (or changing the FP paradigm to a degree of making it globally attractive).
Secondly, if it’s already changed then by the right people in various ways - what the heck does Verse have to offer? I think your insight further drives home the idea that Verse is just a “fun company project”, at least to me!
To be pedantic, generators are from CLU, which is where a lot of languages got them. I’d have to check my research notes but I think pattern matching is from Prolog? Definitely popularized by FP though.
Verse is a domain specific scripting language for UEFN, the Unreal Editor for Fortnite. It’s the language used to describe game logic. The goal is to allow game logic to run in a distributed fashion across many servers, where there could be a hundred thousand players in the same game.
This is a pretty extreme requirement, I don’t actually know how you can reasonably do this in a conventional language. It makes sense that they’ve chosen a logic programming language, because that allows you to express a set of requirements at a high level, without writing a ton of code to specifically take into account the fact that your logic is running in a distributed fashion. According to what I’ve read about Verse, the existing logic languages they looked at have non-deterministic evaluation, which make code unpredictable and hard to debug. So Verse is a deterministic logic language, and that is called out as a specific contribution in the research paper about the Verse calculus.
As for whether Verse will “succeed”. The only job it has to do is to describe game logic in UEFN, and the criteria for success is that it scales to allow very large game environments, in a way that Epic hasn’t succeeded in doing using C++. There’s lots of money available and they’ve hired the best people. so I think they have a good chance of succeeding.
At the GDC talk they mentioned this “BetaVerse” is only temporary, and they intend to have a full specification for a “MaxVerse” which may include many more refined features. It seems like they’re forming the syntax while anticipating complex features like user defined effects and dependent types.
I think Verse will open a new direction for mainstream languages, with dependent types, logic programming and more stuff. I’m not sure if Verse will be popular, but I think Verse will definitely be an influence, both on programmers and designers.
Is there a good place to read up about this language itself. I’m idly curious about their excuses for procrastinating on writing a game by well reasoned design decisions for inventing a new programming language.
Is it natural for me to have skeptical feelings for something like this?… The reasoning being, there has already been so many man hours poured into FP languages, by various experts, and it hasn’t been “paradigm changing”. On top of that, optimizing FP seems to take a lot of work and knowledge. You’re telling me they’ve got all the experts, and hammered something out (which itself sounds inadequate to achieve something great here), which is better than GHC or similar? It’s incredible if true. Like insanely incredible. I would love for someone to explain to me why Verse has a chance at all.
It definitely it has been, look around! Languages have been copying lots of things from FP tools. Algebraic data types, higher-order functions everywhere, pattern matching, immutable by default, futures and promises, generators, etc. These features are many decades old in Haskell, OCaml, etc.
FP and these experts have already changed the paradigm.
No, it’s completely different. Verse is absolutely definitely not Haskell and never will be. The goal of GHC is totally different to what Epic Games is doing with Verse.
People in Epic Games have been thinking about this for decades, recently invested a bunch of time, hired a bunch of experts, made a language, decided to rewrite one of the most popular games in the world using the language they just made. So yeah, I think “Verse has a chance” at doing what it was exactly designed to do. Why not?
Verse’s current implementation compiles to Unreal Engine’s Blueprint VM, so I imagine the integration into Fortnite was simple and not a complete rewrite. The
/Fortnite.com/package contains a bunch of Fornite specific APIs that FII out to Fortnite code iiucYes, the initial integration uses existing tools, but from Tim Sweeney (just after 33 minute mark):
I really hope the Open™ implementation is implemented on top of something a lot more efficient. Blueprint is horribly wasteful. The VM and bytecode has been largely the same since the UnrealScript days - for the curious, it’s essentially a glorified tree-walk interpreter (but with compact bytecode instead of heap-allocated nodes.) It really makes me wonder why Epic decided to continue using the same inefficient tech for UE4’s Blueprint.
The guy who presented the language on stage built a lot of JavaScriptCore / WebKit internals, so they certainly have the right people to build a very fast VM.
I should’ve been clearer in what I meant, but this is definitely true! Definitely true… What I meant was any particular FP language hooking everyone into the FP paradigm (or changing the FP paradigm to a degree of making it globally attractive).
Secondly, if it’s already changed then by the right people in various ways - what the heck does Verse have to offer? I think your insight further drives home the idea that Verse is just a “fun company project”, at least to me!
Thanks for the comment :)
To be pedantic, generators are from CLU, which is where a lot of languages got them. I’d have to check my research notes but I think pattern matching is from Prolog? Definitely popularized by FP though.
Verse is a domain specific scripting language for UEFN, the Unreal Editor for Fortnite. It’s the language used to describe game logic. The goal is to allow game logic to run in a distributed fashion across many servers, where there could be a hundred thousand players in the same game.
This is a pretty extreme requirement, I don’t actually know how you can reasonably do this in a conventional language. It makes sense that they’ve chosen a logic programming language, because that allows you to express a set of requirements at a high level, without writing a ton of code to specifically take into account the fact that your logic is running in a distributed fashion. According to what I’ve read about Verse, the existing logic languages they looked at have non-deterministic evaluation, which make code unpredictable and hard to debug. So Verse is a deterministic logic language, and that is called out as a specific contribution in the research paper about the Verse calculus.
As for whether Verse will “succeed”. The only job it has to do is to describe game logic in UEFN, and the criteria for success is that it scales to allow very large game environments, in a way that Epic hasn’t succeeded in doing using C++. There’s lots of money available and they’ve hired the best people. so I think they have a good chance of succeeding.
At the GDC talk they mentioned this “BetaVerse” is only temporary, and they intend to have a full specification for a “MaxVerse” which may include many more refined features. It seems like they’re forming the syntax while anticipating complex features like user defined effects and dependent types.
I think Verse will open a new direction for mainstream languages, with dependent types, logic programming and more stuff. I’m not sure if Verse will be popular, but I think Verse will definitely be an influence, both on programmers and designers.
Is there a good place to read up about this language itself. I’m idly curious about their
excuses for procrastinating on writing a game bywell reasoned design decisions for inventing a new programming language.Here’s the language reference and here’s the talk introducing it at GDC