It seems similar in concept to Haxe, but focused on writing libraries which are then called from the target language, rather than building your whole application in Fusion. Is that correct?
How are exceptions translated to languages which don't have exceptions, like C?
The amount of targets seems impressive, but I don't see anything resembling a standard library. Usually, it's in the standard library where diverging behavior between targets shows up. Haxe's documentation tends to cover those issues, and lets you know when a behavior is target-dependent, or unspecified. They do try to be consistent between targets whenever possible in a performant way.
Loreline[0] is a recent project which leverages Haxe to build a cross-platform library. Is there a similar project showcasing how Fusion can be used for that?
> I've read about Haxe in 2000s, before I started my work on Fusion. They have different design goals: in Haxe you create whole apps, in Fusion you create components to be used from other languages. Haxe has syntax similar to the (now dead) ActionScript, Fusion is similar to C#. Fusion transpiles to C, D, Swift, TypeScript, OpenCL and Haxe does not.
At a time I was very interested in haxe, but their focus on games made it (perceived as) lacking in the area I wanted to use it (cross platform client-server apps). Recently rust seems to have taken this role for me.
It's a nice idea but the C portion expects an array of unknown size, no bounds checking possible at all and it's also not returned how large the PNM image is. I also don't understand why glib is needed here to get the array to work. It's also missing the stdint.h include to be able to use uint8_t.
> implementing reusable components (libraries) for C, C++, C#, D, Java, JavaScript, Python, Swift, TypeScript and OpenCL C, all from single codebase
Why is this needed? I can't imagine that. I am sure writing code in fusion will produce C++ and Python code which is suboptimal and doesn't fit well in these languages.
I think the target application is writing the same algorithm in multiple places with a guarantee that the logic will be based on a single source of truth. Not unlike Protocol Buffers work to standardize data layout across platforms.
It still feels overcomplicated compared to the standard solution of writing a library in a compiled language you like, exposing a C ABI compatible interface, and hooking it up to any language that can work with that (i.e. any language).
I personally am thinking that developing more programming languages has stopped making sense a long time ago because all of them ignore this untapped market that would help more languages.
C based FFI uniquely privileges C the same way the dollar being a global reserve currency privileges the US, but in the case of currencies the differences in how they work are very small, but with programming languages the differences are substantial.
Half the people working on new programming languages should have been working on GObject competitors instead.
One of the things that is difficult to grasp is that the best FFI is not native to a specific programming language, meaning it does not privilege any specific programming language. Of course this doesn't mean that it cannot be heavily inspired by a particular programming language, it just means that the implementation must be common to all programming languages.
What I mean by the latter is that the features need to be implementable in all programming languages and the easiest way to do it is to define common data types.
The big fallacy that the C FFI advocates present is that it is somehow fine to require full C compiler infrastructure and the C standard library to parse C headers and produce the ABI, but somehow it is the height of evil to define a fat pointer as a C struct consisting of a data pointer and a function pointer and mandating every language to use it, because it is not a C native datatype and only the mythical programming language C is allowed to be used for FFI. I'm pretty sure half the battle could have been won by C having defined common datatypes for the sole purpose of ffi in its standard library.
There have been enormous strides here. We have developed WebAssembly and WIT, with Extism as a chrome over then providing tower of babel between most programming language pairs. We have GraalVM and Truffle to unify languages under one compiler. We have LLVM bitcode to combine languages targeting native machine code for many architectures directly. Swift, Zig, Jai all implement C interop on their own terms. AI agents allow us to re-target algorithms between languages with no compiler interoperability. There are countless pretty darn good ways for programs to communicate through message passing - MachO XPC, Dbus, COM, gRPC, HTTP3 (+ , ZeroMQ, Fuchsia Zircon+FIDL, unix domain sockets. There are more programmers now than at any time in history.
You don't need a C ABI here because the language could in theory generate both the library and the cross language bindings, the underlying mechanism then becomes a mere implementation detail.
I feel like there is a huge missed opportunity that Fusion Lang overlooked.
> writing the same algorithm in multiple places with
This seems for me to be reasonable. Like for a video game with server written in Go/C++/Rust and client written in C#/Javascript, where the same calculations should be done on both sides.
ORMs and variations like Protobuf or things that have to be cross-plateform in the wide sense. The perspective that the same source will behave the same in various environments, and "velocity" trumps performance considerations. If you want to work on things where performance matters, consider embedded/firmware programming ;-)
Nice stuff, I'm working on something like Kaitai Struct, and it'll be very convenient to codegen to Fusion, to support multiple target languages.
However, what I like the least in all kind of software is underdocumentation. There's some reference, it's not bad, but actually it's just bunch of examples. If I want to find out exactly how a particular type or statement translates into a specific language, I'll have to dig into the code.
As others say ; how does debugging work? I was a user of Haxe and Monkey and both had/have serious debugging issues; I think this can only work if it has an interpreter for itself (self hosted I guess ideally) with a proper debugger.
Now we need an UnFusion programming language that can take code written in any language and convert it into Fusion. With that, you could transform any project from language A into language B.
Nice idea. I'm just wondering how to debug code written in fusion... probably you must focus on one of outputs, debug that one, and then back-fit the changes to the fusion source. :/
I like it. I just wonder if today transpiling is better then Agentic Programming. Any specific use case where i should use it? With a good model, SKILLS.md, AGENTS.md and prompts I am able to do - allegedly - a better job then a transpiler, no?
The point, AFAICT, is not in using all capabilities of all the target languages. Rather, it's about expressing some narrower class of computations and grafting them seamlessly into the target languages. Think of data formats, parsers, network protocols, stuff like handling and rendering of text, etc.
I suppose you can write various algorithms in it, and have that code natively trsnspiled to different languages, for ease of native interoperability. It's unlikely to produce the absolutely most optimized code, but the lack of the interface translation barrier (aka FFI) may more than compensate for it.
Rust is not easy to target efficiently, due to the borrow checker, and they likely don't want to dyn Box everything.
Fusion is a programming language designed for implementing reusable components (libraries) for C, C++, C#, D, Java, JavaScript, Python, Swift, TypeScript and OpenCL C, all from single codebase.
Bad idea. It either takes a useless subset of all languages or do weird abstraction that wouldn't fit in a language otherwise.
It seems similar in concept to Haxe, but focused on writing libraries which are then called from the target language, rather than building your whole application in Fusion. Is that correct?
How are exceptions translated to languages which don't have exceptions, like C?
The amount of targets seems impressive, but I don't see anything resembling a standard library. Usually, it's in the standard library where diverging behavior between targets shows up. Haxe's documentation tends to cover those issues, and lets you know when a behavior is target-dependent, or unspecified. They do try to be consistent between targets whenever possible in a performant way.
Loreline[0] is a recent project which leverages Haxe to build a cross-platform library. Is there a similar project showcasing how Fusion can be used for that?
[0] - https://loreline.app/en/docs/
I'd love to see a comparison to Haxe. https://haxe.org/
I wonder what performance and generated code size/quality look like.
Apparently Haxe cannot target OpenCL. It can target PHP and Lua instead.
From https://github.com/fusionlanguage/fut/discussions/119#discus...
> I've read about Haxe in 2000s, before I started my work on Fusion. They have different design goals: in Haxe you create whole apps, in Fusion you create components to be used from other languages. Haxe has syntax similar to the (now dead) ActionScript, Fusion is similar to C#. Fusion transpiles to C, D, Swift, TypeScript, OpenCL and Haxe does not.
At a time I was very interested in haxe, but their focus on games made it (perceived as) lacking in the area I wanted to use it (cross platform client-server apps). Recently rust seems to have taken this role for me.
It's a nice idea but the C portion expects an array of unknown size, no bounds checking possible at all and it's also not returned how large the PNM image is. I also don't understand why glib is needed here to get the array to work. It's also missing the stdint.h include to be able to use uint8_t.
We seem to be regressing to the overuse of classes again.
> implementing reusable components (libraries) for C, C++, C#, D, Java, JavaScript, Python, Swift, TypeScript and OpenCL C, all from single codebase
Why is this needed? I can't imagine that. I am sure writing code in fusion will produce C++ and Python code which is suboptimal and doesn't fit well in these languages.
I think the target application is writing the same algorithm in multiple places with a guarantee that the logic will be based on a single source of truth. Not unlike Protocol Buffers work to standardize data layout across platforms.
It still feels overcomplicated compared to the standard solution of writing a library in a compiled language you like, exposing a C ABI compatible interface, and hooking it up to any language that can work with that (i.e. any language).
C code called from other languages has terrible ergonomics. Writing it sucks, debugging it sucks, maintaining it sucks.
I don't know if fusion is the solution, but I know C isn't.
It can suck but it doesn't always. It depends a lot on the calling language.
What about a better ABI abstraction?
There is GObject, COM, C++/CX and WinRT.
I personally am thinking that developing more programming languages has stopped making sense a long time ago because all of them ignore this untapped market that would help more languages.
C based FFI uniquely privileges C the same way the dollar being a global reserve currency privileges the US, but in the case of currencies the differences in how they work are very small, but with programming languages the differences are substantial.
Half the people working on new programming languages should have been working on GObject competitors instead.
One of the things that is difficult to grasp is that the best FFI is not native to a specific programming language, meaning it does not privilege any specific programming language. Of course this doesn't mean that it cannot be heavily inspired by a particular programming language, it just means that the implementation must be common to all programming languages.
What I mean by the latter is that the features need to be implementable in all programming languages and the easiest way to do it is to define common data types.
The big fallacy that the C FFI advocates present is that it is somehow fine to require full C compiler infrastructure and the C standard library to parse C headers and produce the ABI, but somehow it is the height of evil to define a fat pointer as a C struct consisting of a data pointer and a function pointer and mandating every language to use it, because it is not a C native datatype and only the mythical programming language C is allowed to be used for FFI. I'm pretty sure half the battle could have been won by C having defined common datatypes for the sole purpose of ffi in its standard library.
Exactly! It's surprising to me how little development is happening in this area.
There have been enormous strides here. We have developed WebAssembly and WIT, with Extism as a chrome over then providing tower of babel between most programming language pairs. We have GraalVM and Truffle to unify languages under one compiler. We have LLVM bitcode to combine languages targeting native machine code for many architectures directly. Swift, Zig, Jai all implement C interop on their own terms. AI agents allow us to re-target algorithms between languages with no compiler interoperability. There are countless pretty darn good ways for programs to communicate through message passing - MachO XPC, Dbus, COM, gRPC, HTTP3 (+ , ZeroMQ, Fuchsia Zircon+FIDL, unix domain sockets. There are more programmers now than at any time in history.
You don't need a C ABI here because the language could in theory generate both the library and the cross language bindings, the underlying mechanism then becomes a mere implementation detail.
I feel like there is a huge missed opportunity that Fusion Lang overlooked.
> writing the same algorithm in multiple places with
This seems for me to be reasonable. Like for a video game with server written in Go/C++/Rust and client written in C#/Javascript, where the same calculations should be done on both sides.
ORMs and variations like Protobuf or things that have to be cross-plateform in the wide sense. The perspective that the same source will behave the same in various environments, and "velocity" trumps performance considerations. If you want to work on things where performance matters, consider embedded/firmware programming ;-)
Nice stuff, I'm working on something like Kaitai Struct, and it'll be very convenient to codegen to Fusion, to support multiple target languages.
However, what I like the least in all kind of software is underdocumentation. There's some reference, it's not bad, but actually it's just bunch of examples. If I want to find out exactly how a particular type or statement translates into a specific language, I'll have to dig into the code.
How (and why) are you targeting OpenCL? It has quite a different programming model
As others say ; how does debugging work? I was a user of Haxe and Monkey and both had/have serious debugging issues; I think this can only work if it has an interpreter for itself (self hosted I guess ideally) with a proper debugger.
Now we need an UnFusion programming language that can take code written in any language and convert it into Fusion. With that, you could transform any project from language A into language B.
Nice idea. I'm just wondering how to debug code written in fusion... probably you must focus on one of outputs, debug that one, and then back-fit the changes to the fusion source. :/
It’s not a problem in practice if debug data is emitted properly. For example I used the VS Code C++ debugger extension with Zig.
The syntax is so warm and simple, but not too simple. I like it, but I wonder how debuggable it can be, or if there is an interpreter for learning.
but not Haskell, guess they gave up on picking a string type =P
I like it. I just wonder if today transpiling is better then Agentic Programming. Any specific use case where i should use it? With a good model, SKILLS.md, AGENTS.md and prompts I am able to do - allegedly - a better job then a transpiler, no?
The idea it's good but hard to make it good. A universal language is hard to optimise for a particular language.
The point, AFAICT, is not in using all capabilities of all the target languages. Rather, it's about expressing some narrower class of computations and grafting them seamlessly into the target languages. Think of data formats, parsers, network protocols, stuff like handling and rendering of text, etc.
Wow, this is really fusion. I like it.
Reminds me of https://xkcd.com/927/
What are the use cases? I am curious why Rust was not targeted.
I suppose you can write various algorithms in it, and have that code natively trsnspiled to different languages, for ease of native interoperability. It's unlikely to produce the absolutely most optimized code, but the lack of the interface translation barrier (aka FFI) may more than compensate for it.
Rust is not easy to target efficiently, due to the borrow checker, and they likely don't want to dyn Box everything.
Fusion is a programming language designed for implementing reusable components (libraries) for C, C++, C#, D, Java, JavaScript, Python, Swift, TypeScript and OpenCL C, all from single codebase.