Compiles Xi and oXi source files down to x86_64 assembly.
The compiler is divided into the following phases:
-------------------------------------------------------------------------------------
Usage: xic <OPTION>* <OPERATION>* <FILE>+
-------------------------------------------------------------------------------------
Where <OPTION> is zero or more of:
--help : Print synopsis of options
--report-opts : Print synopsis of optimizations available
-D <DIRECTORY> : Output diagnostic files to <DIRECTORY>
-d <DIRECTORY> : Output assembly files to <DIRECTORY>
-libpath <DIRECTORY> : Search for interface files in <DIRECTORY>
-sourcepath <DIRECTORY> : Search for source files in <DIRECTORY>
-target <OS> : Specify the OS for which to generate code
--optir <PHASE> : Generate .ir file for phase <PHASE>
--optcfg <PHASE> : Generate .dot file for phase <PHASE>
-O<OPT> : Enable optimization <OPT>
-O-no-<OPT> : Disable optimization <OPT>
-O : Disable optimizations, redundant if -O<OPT> passed
-------------------------------------------------------------------------------------
Where <OPERATION> is one or more of:
--lex : For each f.(i)xi, generate lex diagnostic file f.lexed
--parse : For each f.(i)xi, generate parse diagnostic f.(i)parsed
--typecheck : For each f.xi, generate type check diagnostic f.typed
--irgen : For each f.xi, generate intermediate representation f.ir
--irrun : For each f.xi, generate and run IR f.ir
-------------------------------------------------------------------------------------
Where <FILE> is one or more source files to operate on
-------------------------------------------------------------------------------------
Where <PHASE> is exactly one of:
initial : Before any optimizations
cf : Constant folding
cp : Constant propagation
cse : Common subexpression elimination
reg : Register allocation
mc : Move coalescing
final : After all optimizations
-------------------------------------------------------------------------------------
Where <OPT> is exactly one of:
cf : Constant folding
cp : Constant propagation
cse : Common subexpression elimination
reg : Register allocation
mc : Move coalescing
-------------------------------------------------------------------------------------We use the Gradle build system to automate things like lexer and parser generation, compilation, and
executable creation. Run ./xic-build to generate the xic executable.
// hello_world.xi
use io
main(args: int[][]) {
println("Hello, world!")
}Running ./xic -libpath tests/libs hello_world.xi generates:
# hello_world.s
.globl _Imain_paai
.align 4
_Imain_paai:
pushq %rbp
movq %rsp, %rbp
subq $32, %rsp
movq $112, %rdi
callq _xi_alloc
movq %rax, %rcx
movq $13, (%rcx)
movq $72, 8(%rax)
movq $101, 16(%rax)
movq $108, 24(%rax)
movq $108, 32(%rax)
movq $111, 40(%rax)
movq $44, 48(%rax)
movq $32, 56(%rax)
movq $119, 64(%rax)
movq $111, 72(%rax)
movq $114, 80(%rax)
movq $108, 88(%rax)
movq $100, 96(%rax)
movq $33, 104(%rax)
addq $8, %rax
movq %rax, %rdi
callq _Iprintln_pai
jmp _RET_Imain_paai
_RET_Imain_paai:
addq $32, %rsp
popq %rbp
retq(and some library functions).
On a Unix system, we can link this assembly against the provided runtime with
gcc -Lruntime -lxi -lpthread -o hello hello_world.s
And running ./hello prints the expected Hello, world!.