Interpreter Compiler Source Program Source Program Language Processors COMS W4115 Input Interpreter Output Compiler Prof. Stephen A. Edwards Fall 2004 Columbia University Input Executable Program Output Department of Computer Science Bytecode Interpreter Just-in-time Compiler Language Speeds Compared Language Impl. C gcc Source Program Ocaml ocaml SML mlton Source Program C++ g++ SML smlnj Common Lisp cmucl Scheme bigloo Ocaml ocamlb Compiler Java java Pike pike Forth gforth Lua lua Compiler Python python Perl perl Ruby ruby Bytecode Eiffel se Mercury mercury Awk mawk Haskell ghc Lisp rep Icon icon Bytecode Tcl tcl Javascript njs Scheme guile Just-in-time Compiler Forth bigforth Erlang erlang Awk gawk Input Output Emacs Lisp xemacs Scheme stalin Input Bytecode Interpreter Output PHP php Machine Code Bash bash bytecodes native code JIT Threaded code http://www.bagley.org/˜doug/shootout/ Separate Compilation Preprocessor The C Preprocessor foo.c bar.c “Massages” the input before the compiler sees it. cc -E example.c gives #include <stdio.h> extern int C compiler cc: • Macro expansion #define min(x, y) \ printf(char*,...); ((x)<(y))?(x):(y) ... many more declarations • foo.s bar.s printf.o fopen.o malloc.o · · · File inclusion #ifdef DEFINE_BAZ from stdio.h • Conditional compilation int baz(); Assembler as: #endif Archiver ar: void foo() void foo() · · · foo.o bar.o libc.a { { int a = 1; int a = 1; int b = 2; int b = 2; Linker ld: int c; int c; foo — An Executable c = min(a,b); c = ((a)<(b))?(a):(b); } } Compiling a Simple Program What the Compiler Sees Lexical Analysis Gives Tokens int gcd(int a, int b) int gcd(int a, int b) int gcd(int a, int b) { { while (a != b) { while (a != b) { { if (a > b) a -= b; if (a > b) a -= b; while (a != b) { else b -= a; else b -= a; } } if (a > b) a -= b; return a; return a; } } else b -= a; i n t sp g c d ( i n t sp a , sp i int gcd ( int a , int b ) { } n t sp b ) nl { nl sp sp w h i l e sp return a; ( a sp ! = sp b ) sp { nl sp sp sp sp i while ( a != b ) { if ( a } f sp ( a sp > sp b ) sp a sp - = sp b ; nl sp sp sp sp e l s e sp b sp - = sp > b ) a -= b ; else b -= a a ; nl sp sp } nl sp sp r e t u r n sp ; } return a ; } a ; nl } nl A stream of tokens. Whitespace, comments removed. Text file is a sequence of characters Parsing Gives an AST Semantic Analysis Resolves Translation into 3-Address Code Symbols L0: sne $1, a, b func func seq $0, $1, 0 int gcd args seq int gcd args seq btrue $0, L1 % while (a != b) arg arg while return sl $3, b, a arg arg while return seq $2, $3, 0 != int a int b if a int a int b != if a btrue $2, L4 % if (a < b) int gcd(int a, int b) a b > -= -= sub a, a, b % a -= b { Symbol a b > -= -= int gcd(int a, int b) while (a != b) { jmp L5 { a b a b b a while (a != b) { if (a > b) a -= b; Table: a b a b b a else b -= a; L4: sub b, b, a % b -= a if (a > b) a -= b; } int a else b -= a; return a; L5: jmp L0 } } return a; int b L1: ret a } Abstract syntax tree built from parsing rules. Types checked; references to symbols resolved Idealized assembly language w/ infinite registers Generation of 80386 Assembly gcd: pushl %ebp % Save FP movl %esp,%ebp movl 8(%ebp),%eax % Load a from stack movl 12(%ebp),%edx % Load b from stack .L8: cmpl %edx,%eax je .L3 % while (a != b) jle .L5 % if (a < b) subl %edx,%eax % a -= b jmp .L8 .L5: subl %eax,%edx % b -= a jmp .L8 .L3: leave % Restore SP, BP ret.