made puts wor for strings as well an some other changes

a.sl

@@ -1,5 +1,5 @@
 import stdlib.sl
 
 : main
-    "hello world" puts_str
+    "hello world" puts
 ;

build/a.asm

@@ -19,6 +19,36 @@ _start:
     mov rax, 60
     syscall
 word_puts:
+	; detects string if top is len>=0 and next is a pointer in [data_start, data_end)
+	mov rax, [r12]      ; len or int value
+	mov rbx, [r12 + 8]  ; possible address
+	cmp rax, 0
+	jl puts_print_int
+	lea r8, [rel data_start]
+	lea r9, [rel data_end]
+	cmp rbx, r8
+	jl puts_print_int
+	cmp rbx, r9
+	jge puts_print_int
+	; treat as string: (addr below len)
+	mov rdx, rax        ; len
+	mov rsi, rbx        ; addr
+	add r12, 16         ; pop len + addr
+	test rdx, rdx
+	jz puts_str_newline_only
+	mov rax, 1
+	mov rdi, 1
+	syscall
+puts_str_newline_only:
+	mov byte [rel print_buf], 10
+	mov rax, 1
+	mov rdi, 1
+	lea rsi, [rel print_buf]
+	mov rdx, 1
+	syscall
+	ret
+
+puts_print_int:
 	mov rax, [r12]
 	add r12, 8
 	mov rbx, rax
@@ -65,25 +95,6 @@ puts_finish_digits:
 	mov rsi, r9
 	syscall
     ret
-word_puts_str:
-	; expects (addr, len) on data stack
-	mov rdx, [r12]
-	add r12, 8
-	mov rsi, [r12]
-	add r12, 8
-	cmp rdx, 0
-	je puts_str_write_newline
-	mov rax, 1
-	mov rdi, 1
-	syscall
-puts_str_write_newline:
-	mov byte [rel print_buf], 10
-	mov rax, 1
-	mov rdi, 1
-	lea rsi, [rel print_buf]
-	mov rdx, 1
-	syscall
-    ret
 word_dup:
 	mov rax, [r12]
 	sub r12, 8
@@ -267,11 +278,13 @@ word_main:
     ; push 11
     sub r12, 8
     mov qword [r12], 11
-    call word_puts_str
+    call word_puts
     ret
 section .data
+data_start:
 str_0: db 104, 101, 108, 108, 111, 32, 119, 111, 114, 108, 100, 0
 str_0_len equ 11
+data_end:
 section .bss
 align 16
 dstack: resb DSTK_BYTES

build/call_syntax_parens.asm

@@ -19,6 +19,36 @@ _start:
     mov rax, 60
     syscall
 word_puts:
+	; detects string if top is len>=0 and next is a pointer in [data_start, data_end)
+	mov rax, [r12]      ; len or int value
+	mov rbx, [r12 + 8]  ; possible address
+	cmp rax, 0
+	jl puts_print_int
+	lea r8, [rel data_start]
+	lea r9, [rel data_end]
+	cmp rbx, r8
+	jl puts_print_int
+	cmp rbx, r9
+	jge puts_print_int
+	; treat as string: (addr below len)
+	mov rdx, rax        ; len
+	mov rsi, rbx        ; addr
+	add r12, 16         ; pop len + addr
+	test rdx, rdx
+	jz puts_str_newline_only
+	mov rax, 1
+	mov rdi, 1
+	syscall
+puts_str_newline_only:
+	mov byte [rel print_buf], 10
+	mov rax, 1
+	mov rdi, 1
+	lea rsi, [rel print_buf]
+	mov rdx, 1
+	syscall
+	ret
+
+puts_print_int:
 	mov rax, [r12]
 	add r12, 8
 	mov rbx, rax
@@ -65,25 +95,6 @@ puts_finish_digits:
 	mov rsi, r9
 	syscall
     ret
-word_puts_str:
-	; expects (addr, len) on data stack
-	mov rdx, [r12]
-	add r12, 8
-	mov rsi, [r12]
-	add r12, 8
-	cmp rdx, 0
-	je puts_str_write_newline
-	mov rax, 1
-	mov rdi, 1
-	syscall
-puts_str_write_newline:
-	mov byte [rel print_buf], 10
-	mov rax, 1
-	mov rdi, 1
-	lea rsi, [rel print_buf]
-	mov rdx, 1
-	syscall
-    ret
 word_dup:
 	mov rax, [r12]
 	sub r12, 8
@@ -296,6 +307,9 @@ word_foo:
     call word_rdrop
     call word_rdrop
     ret
+section .data
+data_start:
+data_end:
 section .bss
 align 16
 dstack: resb DSTK_BYTES

build/loops_and_cmp.asm

@@ -19,6 +19,36 @@ _start:
     mov rax, 60
     syscall
 word_puts:
+	; detects string if top is len>=0 and next is a pointer in [data_start, data_end)
+	mov rax, [r12]      ; len or int value
+	mov rbx, [r12 + 8]  ; possible address
+	cmp rax, 0
+	jl puts_print_int
+	lea r8, [rel data_start]
+	lea r9, [rel data_end]
+	cmp rbx, r8
+	jl puts_print_int
+	cmp rbx, r9
+	jge puts_print_int
+	; treat as string: (addr below len)
+	mov rdx, rax        ; len
+	mov rsi, rbx        ; addr
+	add r12, 16         ; pop len + addr
+	test rdx, rdx
+	jz puts_str_newline_only
+	mov rax, 1
+	mov rdi, 1
+	syscall
+puts_str_newline_only:
+	mov byte [rel print_buf], 10
+	mov rax, 1
+	mov rdi, 1
+	lea rsi, [rel print_buf]
+	mov rdx, 1
+	syscall
+	ret
+
+puts_print_int:
 	mov rax, [r12]
 	add r12, 8
 	mov rbx, rax
@@ -65,25 +95,6 @@ puts_finish_digits:
 	mov rsi, r9
 	syscall
     ret
-word_puts_str:
-	; expects (addr, len) on data stack
-	mov rdx, [r12]
-	add r12, 8
-	mov rsi, [r12]
-	add r12, 8
-	cmp rdx, 0
-	je puts_str_write_newline
-	mov rax, 1
-	mov rdi, 1
-	syscall
-puts_str_write_newline:
-	mov byte [rel print_buf], 10
-	mov rax, 1
-	mov rdi, 1
-	lea rsi, [rel print_buf]
-	mov rdx, 1
-	syscall
-    ret
 word_dup:
 	mov rax, [r12]
 	sub r12, 8
@@ -305,6 +316,9 @@ L_for_end_1:
     sub r12, 8
     mov qword [r12], 0
     ret
+section .data
+data_start:
+data_end:
 section .bss
 align 16
 dstack: resb DSTK_BYTES

build/main.asm

@@ -19,6 +19,36 @@ _start:
     mov rax, 60
     syscall
 word_puts:
+	; detects string if top is len>=0 and next is a pointer in [data_start, data_end)
+	mov rax, [r12]      ; len or int value
+	mov rbx, [r12 + 8]  ; possible address
+	cmp rax, 0
+	jl puts_print_int
+	lea r8, [rel data_start]
+	lea r9, [rel data_end]
+	cmp rbx, r8
+	jl puts_print_int
+	cmp rbx, r9
+	jge puts_print_int
+	; treat as string: (addr below len)
+	mov rdx, rax        ; len
+	mov rsi, rbx        ; addr
+	add r12, 16         ; pop len + addr
+	test rdx, rdx
+	jz puts_str_newline_only
+	mov rax, 1
+	mov rdi, 1
+	syscall
+puts_str_newline_only:
+	mov byte [rel print_buf], 10
+	mov rax, 1
+	mov rdi, 1
+	lea rsi, [rel print_buf]
+	mov rdx, 1
+	syscall
+	ret
+
+puts_print_int:
 	mov rax, [r12]
 	add r12, 8
 	mov rbx, rax
@@ -277,6 +307,9 @@ word_foo:
     call word_rdrop
     call word_rdrop
     ret
+section .data
+data_start:
+data_end:
 section .bss
 align 16
 dstack: resb DSTK_BYTES

build/override_dup_compile_time.asm

@@ -19,6 +19,36 @@ _start:
     mov rax, 60
     syscall
 word_puts:
+	; detects string if top is len>=0 and next is a pointer in [data_start, data_end)
+	mov rax, [r12]      ; len or int value
+	mov rbx, [r12 + 8]  ; possible address
+	cmp rax, 0
+	jl puts_print_int
+	lea r8, [rel data_start]
+	lea r9, [rel data_end]
+	cmp rbx, r8
+	jl puts_print_int
+	cmp rbx, r9
+	jge puts_print_int
+	; treat as string: (addr below len)
+	mov rdx, rax        ; len
+	mov rsi, rbx        ; addr
+	add r12, 16         ; pop len + addr
+	test rdx, rdx
+	jz puts_str_newline_only
+	mov rax, 1
+	mov rdi, 1
+	syscall
+puts_str_newline_only:
+	mov byte [rel print_buf], 10
+	mov rax, 1
+	mov rdi, 1
+	lea rsi, [rel print_buf]
+	mov rdx, 1
+	syscall
+	ret
+
+puts_print_int:
 	mov rax, [r12]
 	add r12, 8
 	mov rbx, rax
@@ -65,25 +95,6 @@ puts_finish_digits:
 	mov rsi, r9
 	syscall
     ret
-word_puts_str:
-	; expects (addr, len) on data stack
-	mov rdx, [r12]
-	add r12, 8
-	mov rsi, [r12]
-	add r12, 8
-	cmp rdx, 0
-	je puts_str_write_newline
-	mov rax, 1
-	mov rdi, 1
-	syscall
-puts_str_write_newline:
-	mov byte [rel print_buf], 10
-	mov rax, 1
-	mov rdi, 1
-	lea rsi, [rel print_buf]
-	mov rdx, 1
-	syscall
-    ret
 word_dup:
 	mov rax, [r12]
 	sub r12, 8
@@ -269,6 +280,9 @@ word_main:
     sub r12, 8
     mov qword [r12], 0
     ret
+section .data
+data_start:
+data_end:
 section .bss
 align 16
 dstack: resb DSTK_BYTES

build/string_puts.asm

@@ -19,6 +19,36 @@ _start:
     mov rax, 60
     syscall
 word_puts:
+	; detects string if top is len>=0 and next is a pointer in [data_start, data_end)
+	mov rax, [r12]      ; len or int value
+	mov rbx, [r12 + 8]  ; possible address
+	cmp rax, 0
+	jl puts_print_int
+	lea r8, [rel data_start]
+	lea r9, [rel data_end]
+	cmp rbx, r8
+	jl puts_print_int
+	cmp rbx, r9
+	jge puts_print_int
+	; treat as string: (addr below len)
+	mov rdx, rax        ; len
+	mov rsi, rbx        ; addr
+	add r12, 16         ; pop len + addr
+	test rdx, rdx
+	jz puts_str_newline_only
+	mov rax, 1
+	mov rdi, 1
+	syscall
+puts_str_newline_only:
+	mov byte [rel print_buf], 10
+	mov rax, 1
+	mov rdi, 1
+	lea rsi, [rel print_buf]
+	mov rdx, 1
+	syscall
+	ret
+
+puts_print_int:
 	mov rax, [r12]
 	add r12, 8
 	mov rbx, rax
@@ -65,25 +95,6 @@ puts_finish_digits:
 	mov rsi, r9
 	syscall
     ret
-word_puts_str:
-	; expects (addr, len) on data stack
-	mov rdx, [r12]
-	add r12, 8
-	mov rsi, [r12]
-	add r12, 8
-	cmp rdx, 0
-	je puts_str_write_newline
-	mov rax, 1
-	mov rdi, 1
-	syscall
-puts_str_write_newline:
-	mov byte [rel print_buf], 10
-	mov rax, 1
-	mov rdi, 1
-	lea rsi, [rel print_buf]
-	mov rdx, 1
-	syscall
-    ret
 word_dup:
 	mov rax, [r12]
 	sub r12, 8
@@ -267,32 +278,34 @@ word_main:
     ; push 11
     sub r12, 8
     mov qword [r12], 11
-    call word_puts_str
+    call word_puts
     ; push str_1
     sub r12, 8
     mov qword [r12], str_1
     ; push 11
     sub r12, 8
     mov qword [r12], 11
-    call word_puts_str
+    call word_puts
     ; push str_2
     sub r12, 8
     mov qword [r12], str_2
     ; push 0
     sub r12, 8
     mov qword [r12], 0
-    call word_puts_str
+    call word_puts
     ; push 0
     sub r12, 8
     mov qword [r12], 0
     ret
 section .data
+data_start:
 str_0: db 104, 101, 108, 108, 111, 32, 119, 111, 114, 108, 100, 0
 str_0_len equ 11
 str_1: db 108, 105, 110, 101, 49, 10, 108, 105, 110, 101, 50, 0
 str_1_len equ 11
 str_2: db 0
 str_2_len equ 0
+data_end:
 section .bss
 align 16
 dstack: resb DSTK_BYTES

build/test.asm

@@ -19,6 +19,36 @@ _start:
     mov rax, 60
     syscall
 word_puts:
+	; detects string if top is len>=0 and next is a pointer in [data_start, data_end)
+	mov rax, [r12]      ; len or int value
+	mov rbx, [r12 + 8]  ; possible address
+	cmp rax, 0
+	jl puts_print_int
+	lea r8, [rel data_start]
+	lea r9, [rel data_end]
+	cmp rbx, r8
+	jl puts_print_int
+	cmp rbx, r9
+	jge puts_print_int
+	; treat as string: (addr below len)
+	mov rdx, rax        ; len
+	mov rsi, rbx        ; addr
+	add r12, 16         ; pop len + addr
+	test rdx, rdx
+	jz puts_str_newline_only
+	mov rax, 1
+	mov rdi, 1
+	syscall
+puts_str_newline_only:
+	mov byte [rel print_buf], 10
+	mov rax, 1
+	mov rdi, 1
+	lea rsi, [rel print_buf]
+	mov rdx, 1
+	syscall
+	ret
+
+puts_print_int:
 	mov rax, [r12]
 	add r12, 8
 	mov rbx, rax
@@ -730,6 +760,9 @@ word_main:
     sub r12, 8
     mov qword [r12], 0
     ret
+section .data
+data_start:
+data_end:
 section .bss
 align 16
 dstack: resb DSTK_BYTES

hello.sl

@@ -1,5 +1,5 @@
 import stdlib.sl
 
 : main
-    "hello world" puts_str
+    "hello world" puts
 ;

main.py

@@ -1114,13 +1114,25 @@ class Assembler:
 		for definition in runtime_defs:
 			self._emit_definition(definition, emission.text)
 
+		if self._data_section is not None:
+			if not self._data_section:
+				self._data_section.append("data_start:")
+			if not self._data_section or self._data_section[-1] != "data_end:":
+				self._data_section.append("data_end:")
 		emission.bss.extend(self._bss_layout())
 		self._data_section = None
 		return emission
 
+	def _ensure_data_start(self) -> None:
+		if self._data_section is None:
+			raise CompileError("data section is not initialized")
+		if not self._data_section:
+			self._data_section.append("data_start:")
+
 	def _intern_string_literal(self, value: str) -> Tuple[str, int]:
 		if self._data_section is None:
 			raise CompileError("string literal emission requested without data section")
+		self._ensure_data_start()
 		if value in self._string_literals:
 			return self._string_literals[value]
 		label = f"str_{len(self._string_literals)}"

stdlib.sl

@@ -1,4 +1,34 @@
 :asm puts {
+	; detects string if top is len>=0 and next is a pointer in [data_start, data_end)
+	mov rax, [r12]      ; len or int value
+	mov rbx, [r12 + 8]  ; possible address
+	cmp rax, 0
+	jl puts_print_int
+	lea r8, [rel data_start]
+	lea r9, [rel data_end]
+	cmp rbx, r8
+	jl puts_print_int
+	cmp rbx, r9
+	jge puts_print_int
+	; treat as string: (addr below len)
+	mov rdx, rax        ; len
+	mov rsi, rbx        ; addr
+	add r12, 16         ; pop len + addr
+	test rdx, rdx
+	jz puts_str_newline_only
+	mov rax, 1
+	mov rdi, 1
+	syscall
+puts_str_newline_only:
+	mov byte [rel print_buf], 10
+	mov rax, 1
+	mov rdi, 1
+	lea rsi, [rel print_buf]
+	mov rdx, 1
+	syscall
+	ret
+
+puts_print_int:
 	mov rax, [r12]
 	add r12, 8
 	mov rbx, rax
@@ -47,27 +77,6 @@ puts_finish_digits:
 }
 ;
 
-:asm puts_str {
-	; expects (addr, len) on data stack
-	mov rdx, [r12]
-	add r12, 8
-	mov rsi, [r12]
-	add r12, 8
-	cmp rdx, 0
-	je puts_str_write_newline
-	mov rax, 1
-	mov rdi, 1
-	syscall
-puts_str_write_newline:
-	mov byte [rel print_buf], 10
-	mov rax, 1
-	mov rdi, 1
-	lea rsi, [rel print_buf]
-	mov rdx, 1
-	syscall
-}
-;
-
 :asm dup {
 	mov rax, [r12]
 	sub r12, 8

tests/run_tests.py

@@ -105,9 +105,9 @@ compile-only
 import {ROOT / 'stdlib.sl'}
 
 : main
-    \"hello world\" puts_str
+    \"hello world\" puts
-    \"line1\\nline2\" puts_str
+    \"line1\\nline2\" puts
-    \"\" puts_str
+    \"\" puts
     0
 ;
 """,