MLton’s -target flag directs MLton to cross compile a program for
another platform. By default, MLton is only able to compile for the
platform that it is running on. In order to use MLton as a cross
compiler, you need to do two things.
-
Build the MLton runtime system for the target using a remote machine.
-
Install or build the appropriate cross-compiler toolchain on the host so that the C compiler(s) can compile to the target.
To make the terminology clear, we refer to the host as the machine MLton is running on and the target as the machine that MLton is compiling for.
To build the MLton runtime system for the target, use the
remote-add-cross goal of the root Makefile; see comments in
the root Makefile. The remote-add-cross goal uses ssh to
build the runtime system on a remote machine. Building the runtime
system requires compiling and executing programs that determine
characteristics of the platform, which is why it isn’t currently
feasible to build the runtime system on the host machine using a
cross-compiler toolchain.
Here is an example adding a sparc-sun-solaris cross-compile target
using the remote machine blade:
% make CROSS_TARGET=sparc-sun-solaris REMOTE_MACHINE=blade remote-add-cross
Once you have both the MLton runtime system and a cross-compiler toolchain for the target, you should be able to cross compile SML programs. For example,
mlton -target sparc-sun-solaris hello-world.sml
will create hello-world, which you should be able to run on a Sparc
Solaris machine.
Cross-compiling alternatives
Building and maintaining a cross-compiler toolchain is complex. You
may find it simpler to use mlton -keep g to generate the files on
the host, then copy the files to the target, and then use gcc or
mlton on the target to compile the files.