strings

play/car.scm

@@ -0,0 +1 @@
+(prim:write (prim:car (prim:cons 123 456)))

play/cdr.scm

@@ -0,0 +1 @@
+(prim:write (prim:cdr (prim:cons 123 456)))

runtime/src/capi.rs

@@ -0,0 +1,13 @@
+use std::slice;
+
+use crate::scm::scm_bits;
+use crate::scm;
+
+#[unsafe(no_mangle)]
+pub extern "C" fn scm_from_utf8_string (
+    ptr : *const u8,
+    len : usize
+) -> scm_bits {
+    let bytes = unsafe { slice::from_raw_parts (ptr, len) };
+    scm::make_string (str::from_utf8 (bytes).unwrap ())
+}

runtime/src/lib.rs

@@ -1,26 +1,8 @@
 #![allow(non_upper_case_globals)]
 #![allow(non_camel_case_types)]
 
-use std::{io::{stdout, Write}};
 mod gc;
 mod scm;
-use scm::{scm_bits, SCM};
-
-#[unsafe(no_mangle)]
-pub extern "C" fn scm_write (x: scm_bits) -> scm_bits {
-    match scm::unpack (x) {
-        SCM::SmallInt (n) => print! ("{n}"),
-        SCM::Cons (car, cdr) => {
-            print! ("(");
-            scm_write (car);
-            print! (" . ");
-            scm_write (cdr);
-            print! (")");
-        },
-        SCM::Nil => print! ("()"),
-        SCM::False => print! ("#f"),
-        SCM::True => print! ("#t"),
-    };
-    let _ = stdout ().flush ();
-    return 0;
-}
+mod primitives;
+mod obarray;
+mod capi;

runtime/src/obarray.rs

@@ -0,0 +1,10 @@
+use std::{collections::HashMap, hash::{BuildHasher, Hasher}};
+use crate::scm::scm_bits;
+use crate::scm;
+
+mod fnv1a;
+
+pub struct Obarray (HashMap <String, scm_bits>);
+
+pub fn hash () {
+}

runtime/src/obarray/fnv1a.rs

@@ -0,0 +1,28 @@
+use std::{collections::HashMap, hash::{BuildHasher, Hasher}};
+
+pub struct FNV1a (u64);
+pub struct SymbolHash;
+
+impl Hasher for FNV1a {
+    fn finish (&self) -> u64 {
+        self.0
+    }
+
+    fn write (&mut self, bytes: &[u8]) {
+        for b in bytes {
+            self.0 ^= *b as u64;
+            self.0 *= prime;
+        }
+    }
+}
+
+impl BuildHasher for SymbolHash {
+    type Hasher = FNV1a;
+
+    fn build_hasher (&self) -> Self::Hasher {
+        FNV1a (offset_basis)
+    }
+}
+
+const offset_basis : u64 = 0xcbf29ce484222325;
+const prime : u64 = 0x100000001b3;

runtime/src/primitives.rs

@@ -0,0 +1,23 @@
+use crate::scm;
+use crate::scm::{scm_bits, SCM};
+use std::io::{stdout, Write};
+
+#[unsafe(no_mangle)]
+pub extern "C" fn scm_write (x: scm_bits) -> scm_bits {
+    match scm::unpack (x) {
+        SCM::SmallInt (n) => print! ("{n}"),
+        SCM::Cons (car, cdr) => {
+            print! ("(");
+            scm_write (car);
+            print! (" . ");
+            scm_write (cdr);
+            print! (")");
+        },
+        SCM::String (s) => print! ("\"{s}\""),
+        SCM::Nil => print! ("()"),
+        SCM::False => print! ("#f"),
+        SCM::True => print! ("#t"),
+    };
+    let _ = stdout ().flush ();
+    return 0;
+}

runtime/src/scm.rs

@@ -1,6 +1,10 @@
 #![allow(non_upper_case_globals)]
 #![allow(non_camel_case_types)]
 
+use std::slice;
+
+use crate::gc;
+
 pub type scm_bits = u64;
 
 pub const tc2_int      : u64 = 2;
@@ -16,6 +20,7 @@ pub const tc7_string   : u64 = 0x15;
 pub enum SCM {
     SmallInt (i64), 
     Cons (scm_bits, scm_bits),
+    String (String),
     Nil,
     False,
     True,
@@ -35,6 +40,18 @@ pub fn unpack (x : scm_bits) -> SCM {
     } else if is_cons (x) {
         // `car` x and `cdr` x are safe iff `is_cons` x.
         unsafe { SCM::Cons (car (x), cdr (x)) }
+    } else if is_string (x) {
+        let len = unsafe { cell_word (x, 1) };
+        let str_beginning = (x as *const scm_bits).wrapping_add (2) as *const u8;
+        let slice = unsafe {
+            str::from_utf8 (
+                slice::from_raw_parts (
+                    str_beginning,
+                    len.try_into ().unwrap ()
+                )
+            ).unwrap ()
+        };
+        return SCM::String (String::from (slice));
     } else {
         // concat_panic! ("don't know how to unpack: ", x)
         panic! ("don't know how to unpack {x:#016x}")
@@ -49,6 +66,10 @@ const fn is_immediate (x: scm_bits) -> bool {
     6 & x != 0
 }
 
+fn is_string (x: scm_bits) -> bool {
+    has_tc7 (x, tc7_string)
+}
+
 fn is_cons (x: scm_bits) -> bool {
     // safety of `cell_type` is mutually exclusive with
     // `is_immediate`, so this is okay.
@@ -57,6 +78,12 @@ fn is_cons (x: scm_bits) -> bool {
     }
 }
 
+fn has_tc7 (x: scm_bits, tc7: u64) -> bool {
+    unsafe {
+        ! is_immediate (x) && (0x7f & cell_type (x)) == tc7
+    }
+}
+
 unsafe fn cell_type (x: scm_bits) -> scm_bits {
     unsafe { cell_word (x, 0) }
 }
@@ -75,3 +102,50 @@ unsafe fn car (x: scm_bits) -> scm_bits {
 unsafe fn cdr (x: scm_bits) -> scm_bits {
     unsafe { cell_word (x, 1) }
 }
+
+pub unsafe fn words (tag : scm_bits, n : usize) -> *mut scm_bits {
+    let r = unsafe { gc::malloc (n * size_of::<scm_bits> ()) };
+    unsafe { *r = tag };
+    return r
+}
+
+pub fn pack_ptr (obj : *const scm_bits) -> scm_bits {
+    obj as scm_bits
+}
+
+pub unsafe fn set_word (obj : *mut scm_bits, ix : usize, val : scm_bits) {
+    let x = obj.wrapping_add (ix);
+    unsafe { *x = val; }
+}
+
+
+
+pub fn make_string_from_raw_parts (
+    ptr : *const u8,
+    len : usize
+) -> scm_bits {
+    let bytes = unsafe { slice::from_raw_parts (ptr, len) };
+    make_string (str::from_utf8 (bytes).unwrap ())
+}
+
+pub fn make_string (s : &str) -> scm_bits {
+    let len = s.len ();
+    let size_of_tag_and_len = 2 * size_of::<scm_bits> ();
+    let size_of_contents = len;
+    let r = unsafe { gc::malloc (size_of_tag_and_len + size_of_contents) };
+    unsafe {
+        set_word (r, 0, tc7_string);
+        set_word (r, 1, len as u64);
+    }
+    let str_beginning = r.wrapping_add (2) as *mut u8;
+    for (i, b) in s.as_bytes ().iter ().enumerate () {
+        unsafe { *(str_beginning.wrapping_add (i)) = *b };
+    }
+    return pack_ptr (r)
+}
+
+
+
+pub fn make_symbol (name : &str) -> scm_bits {
+    todo! ()
+}

runtime/src/weak-set.rs

@@ -1,2 +0,0 @@
-use std::collections;
-