basic programs can now execute correctly

include/lang2/bytecode.h

@@ -46,11 +46,14 @@ enum l2_opcode {
 
 	/*
 	 * Call a function.
-	 * Pop <word>
+	 * Pop <argc>
+	 * Pop <func>
+	 * Pop argc times
 	 * Push <iptr> + 1
-	 * Jump to <word>
+	 * Push array with args
+	 * Call <func>
 	 */
-	L2_OP_CALL,
+	L2_OP_FUNC_CALL,
 
 	/*
 	 * Jump relative.
@@ -59,13 +62,6 @@ enum l2_opcode {
 	 */
 	L2_OP_RJMP,
 
-	/*
-	 * Generate a stack frame.
-	 * Alloc namespace <var>
-	 * NSPush <var>
-	 */
-	L2_OP_GEN_STACK_FRAME,
-
 	/*
 	 * Look up a value from the current stack frame.
 	 * Pop <word>
@@ -85,6 +81,7 @@ enum l2_opcode {
 	/*
 	 * Return from a function.
 	 * NSPop
+	 * Pop (discard args array)
 	 * Pop <word>
 	 * Jump to <word>
 	 */

include/lang2/gen/gen.h

@@ -23,14 +23,15 @@ void l2_gen_flush(struct l2_generator *gen);
 void l2_gen_free(struct l2_generator *gen);
 
 void l2_gen_halt(struct l2_generator *gen);
-void l2_gen_stack_frame(struct l2_generator *gen);
 void l2_gen_rjmp(struct l2_generator *gen, l2_word len);
 void l2_gen_pop(struct l2_generator *gen);
+void l2_gen_push(struct l2_generator *gen, l2_word word);
 void l2_gen_ret(struct l2_generator *gen);
 void l2_gen_assignment(struct l2_generator *gen, char **ident);
 void l2_gen_number(struct l2_generator *gen, double num);
 void l2_gen_string(struct l2_generator *gen, char **str);
 void l2_gen_function(struct l2_generator *gen, l2_word pos);
 void l2_gen_namespace_lookup(struct l2_generator *gen, char **ident);
+void l2_gen_func_call(struct l2_generator *gen);
 
 #endif

include/lang2/parse/lex.h

@@ -13,6 +13,7 @@ enum l2_token_kind {
 	L2_TOK_COMMA,
 	L2_TOK_PERIOD,
 	L2_TOK_COLON_EQ,
+	L2_TOK_EOL,
 	L2_TOK_EOF,
 	L2_TOK_NUMBER,
 	L2_TOK_STRING,
@@ -42,6 +43,7 @@ struct l2_lexer {
 	int tokidx;
 	int line;
 	int ch;
+	int parens;
 
 	struct l2_bufio_reader reader;
 };
@@ -49,5 +51,6 @@ struct l2_lexer {
 void l2_lexer_init(struct l2_lexer *lexer, struct l2_io_reader *r);
 struct l2_token *l2_lexer_peek(struct l2_lexer *lexer, int count);
 void l2_lexer_consume(struct l2_lexer *lexer);
+void l2_lexer_skip_opt(struct l2_lexer *lexer, enum l2_token_kind kind);
 
 #endif

include/lang2/vm/vm.h

@@ -6,6 +6,10 @@
 #include "../bytecode.h"
 #include "../bitset.h"
 
+struct l2_vm;
+struct l2_vm_array;
+typedef l2_word (*l2_vm_cfunction)(struct l2_vm *vm, struct l2_vm_array *args);
+
 struct l2_vm_value {
 	enum l2_value_flags {
 		L2_VAL_TYPE_NONE,
@@ -15,6 +19,7 @@ struct l2_vm_value {
 		L2_VAL_TYPE_ARRAY,
 		L2_VAL_TYPE_NAMESPACE,
 		L2_VAL_TYPE_FUNCTION,
+		L2_VAL_TYPE_CFUNCTION,
 		L2_VAL_MARKED = 1 << 7,
 		L2_VAL_CONST = 1 << 8,
 	} flags;
@@ -25,6 +30,7 @@ struct l2_vm_value {
 			l2_word pos;
 			l2_word namespace;
 		} func;
+		l2_vm_cfunction cfunc;
 		void *data;
 	};
 };

lib/gen/gen.c

@@ -28,10 +28,6 @@ void l2_gen_halt(struct l2_generator *gen) {
 	put(gen, L2_OP_HALT);
 }
 
-void l2_gen_stack_frame(struct l2_generator *gen) {
-	put(gen, L2_OP_GEN_STACK_FRAME);
-}
-
 void l2_gen_rjmp(struct l2_generator *gen, l2_word len) {
 	put(gen, L2_OP_PUSH);
 	put(gen, len);
@@ -42,6 +38,11 @@ void l2_gen_pop(struct l2_generator *gen) {
 	put(gen, L2_OP_POP);
 }
 
+void l2_gen_push(struct l2_generator *gen, l2_word word) {
+	put(gen, L2_OP_PUSH);
+	put(gen, word);
+}
+
 void l2_gen_ret(struct l2_generator *gen) {
 	put(gen, L2_OP_RET);
 }
@@ -111,3 +112,7 @@ void l2_gen_namespace_lookup(struct l2_generator *gen, char **ident) {
 	put(gen, atom_id);
 	put(gen, L2_OP_STACK_FRAME_LOOKUP);
 }
+
+void l2_gen_func_call(struct l2_generator *gen) {
+	put(gen, L2_OP_FUNC_CALL);
+}

lib/parse/lex.c

@@ -40,6 +40,8 @@ const char *l2_token_kind_name(enum l2_token_kind kind) {
 		return "period";
 	case L2_TOK_COLON_EQ:
 		return "period";
+	case L2_TOK_EOL:
+		return "end-of-line";
 	case L2_TOK_EOF:
 		return "end-of-file";
 	case L2_TOK_NUMBER:
@@ -70,6 +72,7 @@ void l2_lexer_init(struct l2_lexer *lexer, struct l2_io_reader *r) {
 	lexer->tokidx = 0;
 	lexer->line = 1;
 	lexer->ch = 1;
+	lexer->parens = 0;
 	l2_bufio_reader_init(&lexer->reader, r);
 }
 
@@ -93,8 +96,16 @@ static int is_whitespace(int ch) {
 	return ch == ' ' || ch == '\r' || ch == '\n' || ch == '\t';
 }
 
-static void skip_whitespace(struct l2_lexer *lexer) {
-	while (is_whitespace(l2_bufio_peek(&lexer->reader, 1))) read_ch(lexer);
+static int skip_whitespace(struct l2_lexer *lexer) {
+	int nl = 0;
+	while (is_whitespace(l2_bufio_peek(&lexer->reader, 1))) {
+		int ch = read_ch(lexer);
+		if (ch == '\n') {
+			nl = 1;
+		}
+	}
+
+	return nl;
 }
 
 static void read_string(struct l2_lexer *lexer, struct l2_token *tok) {
@@ -213,21 +224,27 @@ static void read_ident(struct l2_lexer *lexer, struct l2_token *tok) {
 }
 
 static void read_tok(struct l2_lexer *lexer, struct l2_token *tok) {
-	skip_whitespace(lexer);
-
 	tok->line = lexer->line;
 	tok->ch = lexer->ch;
+	int nl = skip_whitespace(lexer);
+
+	if (nl && lexer->parens == 0) {
+		tok->kind = L2_TOK_EOL;
+		return;
+	}
 
 	int ch = peek_ch(lexer);
 	switch (ch) {
 	case '(':
 		read_ch(lexer);
 		tok->kind = L2_TOK_OPEN_PAREN;
+		lexer->parens += 1;
 		break;
 
 	case ')':
 		read_ch(lexer);
 		tok->kind = L2_TOK_CLOSE_PAREN;
+		lexer->parens -= 1;
 		break;
 
 	case '{':
@@ -317,3 +334,9 @@ void l2_lexer_consume(struct l2_lexer *lexer) {
 	lexer->tokidx -= 1;
 	memmove(lexer->toks, lexer->toks + 1, lexer->tokidx * sizeof(*lexer->toks));
 }
+
+void l2_lexer_skip_opt(struct l2_lexer *lexer, enum l2_token_kind kind) {
+	if (l2_lexer_peek(lexer, 1)->kind == kind) {
+		l2_lexer_consume(lexer);
+	}
+}

lib/parse/parse.c

@@ -1,16 +1,21 @@
 #include "parse/parse.h"
 
-#include <stdbool.h>
-
 #include "gen/gen.h"
 
+static int is_end_tok(struct l2_token *tok) {
+	return
+		tok->kind == L2_TOK_CLOSE_PAREN || tok->kind == L2_TOK_CLOSE_BRACE ||
+		tok->kind == L2_TOK_CLOSE_BRACKET || tok->kind == L2_TOK_EOL ||
+		tok->kind == L2_TOK_EOF;
+}
+
 static int parse_expression(
 		struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err);
 
 static int parse_function_impl(
 		struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
-	l2_gen_stack_frame(gen);
 	l2_lexer_consume(lexer); // {
+	l2_lexer_skip_opt(lexer, L2_TOK_EOL);
 
 	while (1) {
 		struct l2_token *tok = l2_lexer_peek(lexer, 1);
@@ -25,6 +30,8 @@ static int parse_function_impl(
 		if (parse_expression(lexer, gen, err) < 0) {
 			return -1;
 		}
+
+		l2_lexer_skip_opt(lexer, L2_TOK_EOL);
 	}
 
 	l2_gen_ret(gen);
@@ -71,22 +78,24 @@ static int parse_function(
 	return 0;
 }
 
-static int parse_expression(
+static int parse_sub_expression(
 		struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
 	struct l2_token *tok = l2_lexer_peek(lexer, 1);
-	struct l2_token *tok2 = l2_lexer_peek(lexer, 2);
-
-	if (tok->kind == L2_TOK_IDENT && tok2->kind == L2_TOK_COLON_EQ) {
-		char *ident = l2_token_extract_str(tok);
-		l2_lexer_consume(lexer); // ident
-		l2_lexer_consume(lexer); // :=
 
+	if (tok->kind == L2_TOK_OPEN_PAREN) {
+		l2_lexer_consume(lexer); // (
 		if (parse_expression(lexer, gen, err) < 0) {
-			free(ident);
 			return -1;
 		}
 
-		l2_gen_assignment(gen, &ident);
+		tok = l2_lexer_peek(lexer, 1);
+		if (tok->kind != L2_TOK_CLOSE_PAREN) {
+			l2_parse_err(err, tok, "In paren expression: Expected close paren, got %s",
+					l2_token_kind_name(tok->kind));
+			return -1;
+		}
+
+		l2_lexer_consume(lexer); // )
 		return 0;
 	} else if (tok->kind == L2_TOK_NUMBER) {
 		l2_gen_number(gen, tok->v.num);
@@ -106,16 +115,52 @@ static int parse_expression(
 		return parse_function(lexer, gen, err);
 	}
 
-	l2_parse_err(err, tok, "In expression: Unexpected tokens %s, %s",
-			l2_token_kind_name(tok->kind), l2_token_kind_name(tok2->kind));
+	l2_parse_err(err, tok, "In expression: Unexpected token %s",
+			l2_token_kind_name(tok->kind));
 	return -1;
 }
 
-int l2_parse_program(
+static int parse_expression(
 		struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
-	l2_gen_stack_frame(gen);
+	struct l2_token *tok = l2_lexer_peek(lexer, 1);
+	struct l2_token *tok2 = l2_lexer_peek(lexer, 2);
+
+	if (tok->kind == L2_TOK_IDENT && tok2->kind == L2_TOK_COLON_EQ) {
+		char *ident = l2_token_extract_str(tok);
+		l2_lexer_consume(lexer); // ident
+		l2_lexer_consume(lexer); // :=
+
+		if (parse_expression(lexer, gen, err) < 0) {
+			free(ident);
+			return -1;
+		}
+
+		l2_gen_assignment(gen, &ident);
+		return 0;
+	} else if (tok->kind == L2_TOK_IDENT && !is_end_tok(tok2)) {
+		char *ident = l2_token_extract_str(tok);
+		l2_lexer_consume(lexer);
+
+		l2_word count = 0;
+		while (!is_end_tok(l2_lexer_peek(lexer, 1))) {
+			count += 1;
+			if (parse_sub_expression(lexer, gen, err) < 0) {
+				return -1;
+			}
+		}
 
-	bool first = true;
+		l2_gen_push(gen, count);
+		l2_gen_namespace_lookup(gen, &ident);
+		l2_gen_func_call(gen);
+		return 0;
+	}
+
+	return parse_sub_expression(lexer, gen, err);
+}
+
+int l2_parse_program(
+		struct l2_lexer *lexer, struct l2_generator *gen, struct l2_parse_error *err) {
+	int first = 1;
 
 	while (1) {
 		struct l2_token *tok = l2_lexer_peek(lexer, 1);
@@ -134,7 +179,9 @@ int l2_parse_program(
 			return -1;
 		}
 
-		first = false;
+		l2_lexer_skip_opt(lexer, L2_TOK_EOL);
+
+		first = 0;
 	}
 
 	l2_gen_halt(gen);

lib/vm/namespace.c

@@ -91,7 +91,11 @@ static l2_word get(struct l2_vm_namespace *ns, l2_word key) {
 		l2_word hash = (key + i) & ns->mask;
 		l2_word k = ns->data[hash];
 		if (k == 0) {
-			return 0;
+			if (ns->parent == NULL) {
+				return 0;
+			} else {
+				return get(ns->parent->data, key);
+			}
 		} else if (k == key) {
 			return ns->data[ns->size + hash];
 		}

lib/vm/print.c

@@ -137,18 +137,14 @@ void l2_vm_print_op(l2_word *ops, size_t opcount, size_t *ptr) {
 		printf("ADD\n");
 		break;
 
-	case L2_OP_CALL:
-		printf("CALL\n");
+	case L2_OP_FUNC_CALL:
+		printf("FUNC_CALL\n");
 		break;
 
 	case L2_OP_RJMP:
 		printf("RJMP\n");
 		break;
 
-	case L2_OP_GEN_STACK_FRAME:
-		printf("GEN_STACK_FRAME\n");
-		break;
-
 	case L2_OP_STACK_FRAME_LOOKUP:
 		printf("STACK_FRAME_LOOKUP\n");
 		break;

lib/vm/vm.c

@@ -38,8 +38,12 @@ static void gc_mark_array(struct l2_vm *vm, struct l2_vm_value *val);
 static void gc_mark_namespace(struct l2_vm *vm, struct l2_vm_value *val);
 
 static void gc_mark(struct l2_vm *vm, l2_word id) {
-	printf("GC MARK %i\n", id);
 	struct l2_vm_value *val = &vm->values[id];
+	if (val->flags & L2_VAL_MARKED) {
+		return;
+	}
+
+	printf("GC MARK %i\n", id);
 	val->flags |= L2_VAL_MARKED;
 
 	int typ = l2_vm_value_type(val);
@@ -77,6 +81,10 @@ static void gc_mark_namespace(struct l2_vm *vm, struct l2_vm_value *val) {
 
 		gc_mark(vm, ns->data[ns->size + i]);
 	}
+
+	if (ns->parent != NULL) {
+		gc_mark_namespace(vm, ns->parent);
+	}
 }
 
 static void gc_free(struct l2_vm *vm, l2_word id) {
@@ -101,6 +109,7 @@ static size_t gc_sweep(struct l2_vm *vm) {
 
 		struct l2_vm_value *val = &vm->values[i];
 		if (!(val->flags & L2_VAL_MARKED)) {
+			printf("GC FREE %zi\n", i);
 			gc_free(vm, i);
 			freed += 1;
 		} else {
@@ -125,6 +134,13 @@ void l2_vm_init(struct l2_vm *vm, l2_word *ops, size_t opcount) {
 	// variable ID 0 should be the only 'none' variable in the system
 	l2_word none_id = alloc_val(vm);
 	vm->values[none_id].flags = L2_VAL_TYPE_NONE | L2_VAL_CONST;
+
+	// Need to allocate a root namespace
+	l2_word root = alloc_val(vm);
+	vm->values[root].flags = L2_VAL_TYPE_NAMESPACE;
+	vm->values[root].data = NULL; // Will be allocated on first insert
+	vm->nstack[vm->nsptr] = root;
+	vm->nsptr += 1;
 }
 
 void l2_vm_free(struct l2_vm *vm) {
@@ -192,10 +208,51 @@ void l2_vm_step(struct l2_vm *vm) {
 		vm->sptr -= 1;
 		break;
 
-	case L2_OP_CALL:
-		word = vm->stack[vm->sptr - 1];
-		vm->stack[vm->sptr - 1] = vm->iptr + 1;
-		vm->iptr = word;
+	case L2_OP_FUNC_CALL:
+		{
+			l2_word func_id = vm->stack[--vm->sptr];
+			l2_word argc = vm->stack[--vm->sptr];
+			struct l2_vm_value *func = &vm->values[func_id];
+
+			l2_word arr_id = alloc_val(vm);
+			vm->values[arr_id].flags = L2_VAL_TYPE_ARRAY;
+			vm->values[arr_id].data = malloc(
+					sizeof(struct l2_vm_array) + sizeof(l2_word) * argc);
+			struct l2_vm_array *arr = vm->values[arr_id].data;
+			arr->len = argc;
+			arr->size = argc;
+
+			vm->sptr -= argc;
+			for (l2_word i = 0; i < argc; ++i) {
+				arr->data[i] = vm->stack[vm->sptr + i];
+			}
+
+			enum l2_value_flags typ = l2_vm_value_type(func);
+
+			// C functions are called differently from language functions
+			if (typ == L2_VAL_TYPE_CFUNCTION) {
+				vm->stack[vm->sptr++] = func->cfunc(vm, arr);
+				break;
+			}
+
+			// Don't interpret a non-function as a function
+			if (typ != L2_VAL_TYPE_FUNCTION) {
+				// TODO: Error mechanism
+				break;
+			}
+
+			vm->stack[vm->sptr++] = vm->iptr;
+			vm->stack[vm->sptr++] = arr_id;
+
+			l2_word ns_id = alloc_val(vm);
+			vm->values[ns_id].flags = L2_VAL_TYPE_NAMESPACE;
+			vm->values[ns_id].data = calloc(1, sizeof(struct l2_vm_namespace));
+			struct l2_vm_namespace *ns = vm->values[ns_id].data;
+			ns->parent = &vm->values[func->func.namespace]; // TODO: This won't work if values is realloc'd
+			vm->nstack[vm->nsptr++] = ns_id;
+
+			vm->iptr = func->func.pos;
+		}
 		break;
 
 	case L2_OP_RJMP:
@@ -204,14 +261,6 @@ void l2_vm_step(struct l2_vm *vm) {
 		vm->iptr += word;
 		break;
 
-	case L2_OP_GEN_STACK_FRAME:
-		word = alloc_val(vm);
-		vm->values[word].flags = L2_VAL_TYPE_NAMESPACE;
-		vm->values[word].data = NULL; // Will be allocated on first insert
-		vm->nstack[vm->nsptr] = word;
-		vm->nsptr += 1;
-		break;
-
 	case L2_OP_STACK_FRAME_LOOKUP:
 		{
 			l2_word key = vm->stack[vm->sptr - 1];
@@ -231,10 +280,14 @@ void l2_vm_step(struct l2_vm *vm) {
 		break;
 
 	case L2_OP_RET:
-		vm->nsptr -= 1;
-		vm->iptr = vm->stack[vm->sptr - 1];
-		vm->sptr -= 1;
-		vm->nsptr -= 1;
+		{
+			l2_word retval = vm->stack[--vm->sptr];
+			vm->sptr -= 1; // Discard arguments array
+			l2_word retaddr = vm->stack[--vm->sptr];
+			vm->stack[vm->sptr++] = retval;
+			vm->nsptr -= 1; // Pop function stack frame
+			vm->iptr = retaddr;
+		}
 		break;
 
 	case L2_OP_ALLOC_INTEGER_32:

test/src/lex.t.c

@@ -46,6 +46,9 @@ describe(lex) {
 		asserteq(l2_lexer_peek(&lexer, 1)->v.num, 10);
 		l2_lexer_consume(&lexer);
 
+		asserteq(l2_lexer_peek(&lexer, 1)->kind, L2_TOK_EOL);
+		l2_lexer_consume(&lexer);
+
 		asserteq(l2_lexer_peek(&lexer, 1)->kind, L2_TOK_IDENT);
 		asserteq(l2_lexer_peek(&lexer, 1)->v.str, "bar");
 		l2_lexer_consume(&lexer);