removed p! and p@ and added 'mem' that just pushes the addr os the persistent memory on to the stack and than you can write/read using standard ! and @ nd write at an offset uaing pointer arithmetic

alloc.sl

@@ -0,0 +1,36 @@
+import stdlib/stdlib.sl
+import stdlib/io.sl
+
+: alloc
+    0      # addr hint (NULL)
+    swap   # size
+    3      # prot (PROT_READ | PROT_WRITE)
+    34     # flags (MAP_PRIVATE | MAP_ANON)
+    -1     # fd
+    0      # offset
+    mmap
+;
+
+: free
+    munmap drop
+;
+
+: test-mem-alloc
+    4096 alloc dup 1337 swap !   # allocate 4096 bytes, store 1337 at start
+    dup @ puti cr                # print value at start
+    4096 free                    # free the memory
+;
+
+struct: Point
+    field x 8
+    field y 8
+;struct
+
+: main2
+    32 alloc           # allocate 32 bytes (enough for a Point struct)
+    dup 111 swap Point.x!
+    dup 222 swap Point.y!
+    dup Point.x@ puti cr
+    Point.y@ puti cr
+    32 free            # free the memory
+;

main.py

@@ -324,7 +324,6 @@ IntrinsicEmitter = Callable[["FunctionEmitter"], None]
 class Word:
 	name: str
 	immediate: bool = False
-	stack_effect: str = "( -- )"
 	definition: Optional[Union[Definition, AsmDefinition]] = None
 	macro: Optional[MacroHandler] = None
 	intrinsic: Optional[IntrinsicEmitter] = None
@@ -1511,8 +1510,6 @@ class _CTHandleTable:
 class Assembler:
 	def __init__(self, dictionary: Dictionary) -> None:
 		self.dictionary = dictionary
-		self.stack_bytes = 65536
-		self.io_buffer_bytes = 128
 		self._string_literals: Dict[str, Tuple[str, int]] = {}
 		self._float_literals: Dict[float, str] = {}
 		self._data_section: Optional[List[str]] = None

mem.sl

@@ -1,36 +1,8 @@
 import stdlib/stdlib.sl
 import stdlib/io.sl
+import stdlib/debug.sl
 
-: alloc
+: main
-    0      # addr hint (NULL)
+	mem dup 5 swap !
-    swap   # size
+	@ puti cr
-    3      # prot (PROT_READ | PROT_WRITE)
-    34     # flags (MAP_PRIVATE | MAP_ANON)
-    -1     # fd
-    0      # offset
-    mmap
 ;
-
-: free
-    munmap drop
-;
-
-: test-mem-alloc
-    4096 alloc dup 1337 swap !   # allocate 4096 bytes, store 1337 at start
-    dup @ puti cr                # print value at start
-    4096 free                    # free the memory
-;
-
-struct: Point
-    field x 8
-    field y 8
-;struct
-
-: main2
-    32 alloc           # allocate 32 bytes (enough for a Point struct)
-    dup 111 swap Point.x!
-    dup 222 swap Point.y!
-    dup Point.x@ puti cr
-    Point.y@ puti cr
-    32 free            # free the memory
-;

stdlib/stdlib.sl

@@ -1,27 +1,11 @@
 # Reserve 64 bytes in .bss
 # persistent: resb 64
+# push the addr of it
 
-# : p@ ( offset -- n )
+:asm mem {
-:asm p@ {
+	lea rax, [rel persistent]
-	mov rax, [r12]         ; offset
+	sub r12, 8
-	lea rbx, [rel persistent]
+	mov [r12], rax
-	add rax, rbx           ; address = persistent + offset
-	mov rax, [rax]         ; load value
-	mov [r12], rax         ; store on stack
-	ret
-}
-;
-
-# : p! ( value offset -- )
-:asm p! {
-	mov rbx, [r12]         ; offset
-	add r12, 8
-	mov rax, [r12]         ; value
-	lea rcx, [rel persistent]
-	add rcx, rbx           ; address = persistent + offset
-	mov [rcx], rax         ; store value
-	add r12, 8             ; pop value
-	ret
 }
 ;