WebAssembly has a virtual machine architecture where the host is the embedding process and the guest is a program compiled into the WebAssembly Binary Format, also known as Wasm. The first step is to take a source file and compile it into the Wasm bytecode.
Ex. If your source is in Go, you might compile it with TinyGo.
Below are notes wazero contributed so far, in alphabetical order by language.
- Go Ex.
GOARCH=wasm GOOS=js go build -o X.wasm X.go
- TinyGo Ex.
tinygo build -o X.wasm -target=wasi X.go
- Rust Ex.
rustc -o X.wasm --target wasm32-wasi X.rs
wazero is a runtime that embeds in Go applications, not a web browser. As such, these notes bias towards backend use of WebAssembly, not browser use.
Disclaimer: These are not official documentation, nor represent the teams who maintain language compilers. If you see any errors, please help maintain these and star our GitHub repository if they are helpful. Together, we can make WebAssembly easier on the next person.
The WebAssembly Core specification defines a stack-based virtual machine. The only features that work by default are computational in nature, and the only way to communicate is via functions, memory or global variables.
WebAssembly has no standard library or system call interface to implement features the operating system would otherwise provide. Certain capabilities, such as forking a process, will not work. Support of common I/O features, such as writing to the console, vary. See System Calls for more.
Software is more than technical constraints. WebAssembly remains a relatively niche target, with limited maintenance and development. This means that certain features may not work, yet, even if they could technically.
In general, developing with WebAssembly is difficult, and fewer problems can be discovered at compilation time vs more supported targets. This results in more runtime errors, or even panics. Where error messages exist, they may be misleading. Finally, the languages maintainers may be less familiar with how to solve the problems, and/or rely on less available key maintainers.
The above constraints affect the library design and dependency choices in your source, and by extension the choices of library dependencies you can use. In extreme cases, constraints or support concerns may lead developers to choose newer languages like Zig.
Regardless of the programming language used, the best advice is to unit test your code, and run tests with your intended WebAssembly runtime, like wazero.
These tests should cover the critical paths of your code, including errors. Doing so protects your time. You’ll have higher confidence, and more efficient means to communicate problems vs ad-hoc reports.
WebAssembly is a stack-based virtual machine specification, so operates at a lower level than an operating system. For functionality the operating system would otherwise provide, system interfaces are needed.
Programming languages usually include a standard library, with features that
require I/O, such as writing to the console. Portability is helped along with
POSIX conforming implementations of system calls, such as
There is a WebAssembly System Interface, a.k.a. WASI, which defines host functions loosely based on POSIX. There’s also a de facto implementation wasi-libc. However, WASI is not a standard and language compilers don’t always support it.
For example, AssemblyScript once supported WASI, but no longer does. Even
compilers that target WASI using wasi-libc have gaps. For example,
TinyGo does not yet support
fd_readdir. Some toolchains have a
hybrid approach. For example, Emscripten uses WASI for console output, but its
own virtual filesystem functions. Finally, the team behind WASI are
developing an incompatible, modular replacement to the current version.
It is important to note that even when system interfaces are supported, some users prefer a freestanding compilation target that restricts them. This helps them control binary size and performance.
In summary, system interfaces in WebAssembly are not standard and are immature. Developers need to understand and test the system interfaces they rely on. Testing ensures not only the present capabilities, but also they continue to operate as the ecosystem matures.
WebAssembly does not yet support true parallelism; it lacks support for multiple threads, atomics, and memory barriers. (It may someday; See the threads proposal.)
For example, a compiler targeting WASI, generates
_start function corresponding to
main in the original source code. When
the WebAssembly runtime calls
_start, it remains on the same thread of
execution until that function completes.
Concretely, if using wazero, a Wasm function call remains on the calling goroutine until it completes.
In summary, while true that host functions can do anything, including launch processes, Wasm binaries compliant with the [WebAssembly Core Specification] (https://wazero.io/specs/#core) cannot do anything in parallel, unless they use non-standard instructions or conventions not yet defined by the specification.
Compiling Parallel Code to Serial Wasm
Until this changes, language compilers cannot generate Wasm that can control scheduling within a function or safely modify memory in parallel. In other words, one function cannot do anything in parallel.
This impacts how programming language primitives translate to Wasm:
- Garbage collection invokes on the runtime host’s calling thread instead of in the background.
- Language-defined threads or co-routines fail compilation or are limited to sequential processing.
- Locks and barriers fail compilation or are implemented unsafely.
- Async functions including I/O execute sequentially.
Language compilers often used shared infrastructure, such as LLVM and Binaryen. One tool that helps in translation is Binaryen’s Asyncify, which lets a language support synchronous operations in an async manner.
Concurrency via Orchestration
To work around lack of concurrency at the WebAssembly Core abstraction, tools often orchestrate pools of workers, and ensure a module in that pool is only used sequentially.
For example, waPC provides a WASM module pool, so host callbacks can be invoked in parallel, despite not being able to share memory.