From 13f4e2bfe4231b6bc802a1baab40b9165f40c1b2 Mon Sep 17 00:00:00 2001
From: janis <janis@nirgendwo.xyz>
Date: Thu, 11 Jun 2026 17:30:43 +0200
Subject: [PATCH] string parsing/remove nil tag in favour of constant
 pointer/add char tag

---
 stages/lisp0/lisp.asm | 95 +++++++++++++++++++++++++++++++++++--------
 1 file changed, 79 insertions(+), 16 deletions(-)

diff --git a/stages/lisp0/lisp.asm b/stages/lisp0/lisp.asm
index 49e5774..3fe86c7 100644
--- a/stages/lisp0/lisp.asm
+++ b/stages/lisp0/lisp.asm
@@ -98,13 +98,14 @@ extern dealloc
 
     align 8,db 0
     nil dq 1                    ; the nil object, with refcount = 1
-    OBJ_NIL equ 0
+    OBJ_BYTE equ 0
     OBJ_CONS equ 1
     OBJ_CLOS equ 2
     OBJ_ATOM equ 3
-    OBJ_NUM equ 4
+    OBJ_NUM  equ 4
     OBJ_PRIM equ 5
-    OBJ_STR equ 6
+    OBJ_STR  equ 6
+    OBJ_ARR  equ 7
 
     ATOM_QUOTE equ 0
     ATOM_TRUE equ 1
@@ -145,7 +146,7 @@ extern dealloc
     NOT_STR_LEN equ $ - NOT_STR
 
 ;; folds all elements in the list $rdi using the binary function $rdx in the environment $rsi with an accumulator value of $rcx, returning the result in $rax
-    ;; fn fold_list(List<T>, Env, (U, T) -> U, U) -> U
+    ;; fn fold_list(List<T>, Env, (U, T, Env) -> U, U) -> U
 fold_list:
     sub rsp, 32
     mov qword [rsp], rdi        ; rest
@@ -172,7 +173,7 @@ fold_list:
     ret
 
 ;; like fold_list, but requires that the list is non-empty
-    ;; fn reduce_list(List<T>, Env, (T, T) -> T) -> T
+    ;; fn reduce_list(List<T>, Env, (T, T, Env) -> T) -> T
 reduce_list:
     call get_car
     mov rcx, rax                ; acc = car(list)
@@ -187,7 +188,8 @@ reduce_list:
     mov rax, rcx
     ret
 
-    ;; like reduce_list, calls a predicate instead of a binary function, and returns "t" if the predicate returns non-zero for any pair of adjacent elements, and nil otherwise
+;; like reduce_list, calls a predicate instead of a binary function, and returns "t" if the predicate returns non-zero for any pair of adjacent elements, and nil otherwise
+;; predicate: fn (T, T, Env) -> bool
 any2_list:
     sub rsp, 24
     mov qword [rsp], rdi        ; rest
@@ -843,12 +845,12 @@ obj_eq:
 
 
 dtor_table:
-    dd dtor_table - dtor_nil
+    dd 0
     dd dtor_table - dtor_cons
     dd dtor_table - dtor_clos
     dd dtor_table - dtor_atom
     dd dtor_table - dtor_num
-    dd dtor_table - dtor_prim
+    dd 0
     dd dtor_table - dtor_str
     dd 0
 
@@ -883,27 +885,31 @@ dtor_num:
     mov rdx, 8
     call dealloc
     ret
+    
 
 obj_inc_ref:
-    call obj_tag_part
-    cmp al, OBJ_NIL
-    je .done           ; nil is immortal, so we're done
+    call is_nil
+    test al, al
+    jnz .done           ; nil is immortal, so we're done
     call obj_addr_part
     inc dword [rax]    ; increment the refcount
 .done:
     ret
 
 obj_dec_ref:
+    call is_nil
+    test al, al
+    jnz .done           ; nil is immortal, so we're done
     call obj_tag_part
-    cmp al, OBJ_NIL
-    je .done           ; nil is immortal, so we're done
     mov rsi, rax
     call obj_addr_part
     dec dword [rax]    ; decrement the refcount
     jnz .done           ; if refcount != 0, we're done
     lea rcx, qword [rel dtor_table]
     movsxd rsi, [rcx + rsi * 4] ; get the offset of the destructor
-    add rcx, rsi                  ; calculate the address of the destructor
+    test rsi, rsi
+    jz .done                      ; if there's no destructor, we're done
+    add rcx, rsi                  ; p_dtor = &dtor_table + dtor_table[tag]
     mov rdi, rax                  ; set the argument for the destructor (the object pointer)
     call rcx                      ; call the destructor
     .done:
@@ -928,6 +934,18 @@ obj_set_tag:
     or rax, rsi          ; set the new tag bits
     ret
 
+    ;; construcst a closure object with params $rdi, body $rsi, and env $rdx
+clos:
+    push rdx
+    call cons
+    mov rdi, rax
+    pop rsi
+    call cons
+    mov rdi, rax
+    mov rsi, OBJ_CLOS
+    call obj_set_tag
+    ret
+    
     ;; constructs a cons cell ($rdi . $rsi) and returns an object pointer to it
 cons:
     push rdi
@@ -1080,6 +1098,50 @@ parse_cur_token:
     pop rdi
     call parse_quote
     ret
+.string:
+    ret
+
+    ;; fn parse_string(&mut iter) -> Box<[u8]>
+parse_string:
+    push r12
+    xor r12, r12
+.loop:
+    ; scan for closing quote
+    mov al, byte [rel buf + r12]
+    ; cmp al, `\\`
+    ; je .escape
+    cmp al, `"`
+    je .done
+    test al, al
+    jz .invalid
+    inc r12
+    jmp .loop
+.done:
+    mov rdi, r12
+    mov rsi, 1
+    call alloc
+    lea rdi, [rel buf]
+    mov rsi, rax
+    mov rdx, r12
+    call memcpy
+    push rax
+    mov rdi, 16
+    mov rsi, 8
+    call alloc
+    mov dword [rax], 1          ; refcount = 1
+    mov dword [rax + 4], r12w   ; length of string
+    pop rdi
+    mov qword [rax + 8], rdi      ; pointer to string
+    mov rdi, rax
+    mov rsi, OBJ_STR
+    call obj_set_tag
+    pop r12
+    ret
+.invalid:
+    xor rdi, rdi
+    call panic_abort
+
+
 
 parse_num:
     push r12
@@ -1307,8 +1369,9 @@ eval_list:
     mov rdi, qword [rsp + 24]   ; $rdi = cdr($rdi)
     jmp .tailcall
 .not_cons:
-    cmp al, OBJ_NIL
-    je .nil
+    call is_nil
+    test al, al
+    jnz .nil
     mov rsi, qword [rsp]        ; $rsi = environment
     call eval
     mov rsi, rax