It’s assumed that you are familiar with the basics of Rust and the language you will be calling from. You should have read the official FFI documentation, but some basics will be covered here.
[dependencies] libc = "*" [lib] crate-type = ["cdylib"]
crate-type = ["cdylib"] creates a dynamically linked library.
Check out the Cargo documentation for
dynamic or static libraries for more information.
cdylib was introduced in RFC 1510 and improves on the
dylib with reduced file size and fewer exported symbols. It
was implemented in Rust 1.10; if you are using earlier
versions, it is encouraged that you upgrade, but you can also use
dylib with minimal ill effects.
Some examples are so minimal that they don’t use any features from the
Rust standard library. This triggers a known issue and
linking will fail. The only workaround is to include an exported but
unused function that does use something from the standard
library. These functions are called
fix_linking_when_not_using_stdlib and can be safely removed for any
All C examples will be compiled using the C11 standard.
All Ruby examples will use Ruby 2.4 and the FFI gem.
All Python examples will use Python 3.4 and the ctypes library.
All Haskell examples will use GHC 7.6 with the
language extension and only the
base library which comes with GHC.
All Node.js examples will use Node.js 8.7 and the ffi package.
All C# examples will compile with Mono 5.10. It is assumed that this code will work with the Microsoft CLR frameworks, but this is untested.
When running examples, you need to ensure the Rust dynamic library can be located by the system.
With most shells on Mac OS X and Linux, this can be done by prefixing
LD_LIBRARY_PATH=target/debug. For example, to run a
Python example, you might use
LD_LIBRARY_PATH=target/debug python src/main.py from the example
On Windows, the simplest course of action is to copy the compiled
dynamic library into the current working directory before running the
examples. You only need the
.dll file. Also note that when
running Python examples, you may wish to use
py instead of
python, especially if you have multiple versions of Python