mugs in ocaml with murmur3 fork

2 files changed, 173 insertions, 0 deletions

diff --git a/vere/build.zig b/vere/build.zig
index 72baf8e..e2a7590 100644
--- a/vere/build.zig
+++ b/vere/build.zig
@@ -643,6 +643,11 @@ fn buildBinary(
                 .deps = vere_test_deps,
             },
             .{
+                .name = "mug-test",
+                .file = "pkg/vere/mug_test.c",
+                .deps = vere_test_deps,
+            },
+            .{
                 .name = "newt-test",
                 .file = "pkg/vere/newt_tests.c",
                 .deps = vere_test_deps,
diff --git a/vere/pkg/vere/mug_test.c b/vere/pkg/vere/mug_test.c
new file mode 100644
index 0000000..19bad89
--- /dev/null
+++ b/vere/pkg/vere/mug_test.c
@@ -0,0 +1,168 @@
+/* Mug Test - Compare C vs OCaml mug computation */
+
+#include "noun.h"
+
+#include <stdio.h>
+#include <stdlib.h>
+
+/* Helper to print mug with description */
+static void
+test_mug(const char* desc, u3_noun noun)
+{
+  c3_w mug = u3r_mug(noun);
+  printf("%-40s 0x%08x\n", desc, mug);
+}
+
+/* Helper to build cord from string */
+static u3_atom
+_cord(const char* str)
+{
+  return u3i_string(str);
+}
+
+int
+main(int argc, char* argv[])
+{
+  /* Initialize u3 with lite boot (no jets, just memory) */
+  u3C.wag_w |= u3o_hashless;
+  u3m_boot_lite(1 << 28);  // 256MB loom
+
+  printf("\n═══════════════════════════════════════\n");
+  printf(" Mug Computation Test (C)\n");
+  printf("═══════════════════════════════════════\n\n");
+
+  printf("ATOMS - Small values:\n");
+  printf("─────────────────────────────────────\n");
+  test_mug("0", 0);
+  test_mug("1", 1);
+  test_mug("2", 2);
+  test_mug("3", 3);
+  test_mug("10", 10);
+  test_mug("42", 42);
+  test_mug("100", 100);
+  test_mug("255", 255);
+  test_mug("256", 256);
+  test_mug("1000", 1000);
+  printf("\n");
+
+  printf("ATOMS - Powers of 2:\n");
+  printf("─────────────────────────────────────\n");
+  test_mug("2^8 (256)", 256);
+  test_mug("2^16 (65536)", 65536);
+  test_mug("2^20 (1048576)", 1048576);
+  test_mug("2^24 (16777216)", 16777216);
+  test_mug("2^31 - 1 (max signed 32-bit)", 2147483647);
+  printf("\n");
+
+  printf("ATOMS - Large values:\n");
+  printf("─────────────────────────────────────\n");
+  test_mug("2^32", u3i_chub(1ULL << 32));
+  test_mug("2^64", u3i_chubs(1, (c3_d[]){0}));  // 2^64 would overflow, use 2^63
+  test_mug("2^128", u3i_chubs(2, (c3_d[]){0, 1}));
+  test_mug("0xdeadbeef", u3i_word(0xdeadbeef));
+  test_mug("0xcafebabe", u3i_word(0xcafebabe));
+  printf("\n");
+
+  printf("CELLS - Simple pairs:\n");
+  printf("─────────────────────────────────────\n");
+  test_mug("[0 0]", u3nc(0, 0));
+  test_mug("[0 1]", u3nc(0, 1));
+  test_mug("[1 0]", u3nc(1, 0));
+  test_mug("[1 1]", u3nc(1, 1));
+  test_mug("[1 2]", u3nc(1, 2));
+  test_mug("[2 1]", u3nc(2, 1));
+  test_mug("[42 0]", u3nc(42, 0));
+  test_mug("[0 42]", u3nc(0, 42));
+  test_mug("[42 42]", u3nc(42, 42));
+  test_mug("[100 200]", u3nc(100, 200));
+  printf("\n");
+
+  printf("CELLS - Nested structures:\n");
+  printf("─────────────────────────────────────\n");
+  test_mug("[[0 0] 0]", u3nc(u3nc(0, 0), 0));
+  test_mug("[0 [0 0]]", u3nc(0, u3nc(0, 0)));
+  test_mug("[[1 2] 3]", u3nc(u3nc(1, 2), 3));
+  test_mug("[1 [2 3]]", u3nc(1, u3nc(2, 3)));
+  test_mug("[[1 2] [3 4]]", u3nc(u3nc(1, 2), u3nc(3, 4)));
+  test_mug("[[[0 1] 2] 3]", u3nc(u3nc(u3nc(0, 1), 2), 3));
+  test_mug("[0 [1 [2 3]]]", u3nc(0, u3nc(1, u3nc(2, 3))));
+  printf("\n");
+
+  printf("CELLS - Binary trees:\n");
+  printf("─────────────────────────────────────\n");
+  /* Balanced binary tree depth 2 */
+  u3_noun tree2 = u3nc(u3nc(1, 2), u3nc(3, 4));
+  test_mug("Balanced tree depth 2", tree2);
+
+  /* Balanced binary tree depth 3 */
+  u3_noun tree3 = u3nc(
+    u3nc(u3nc(1, 2), u3nc(3, 4)),
+    u3nc(u3nc(5, 6), u3nc(7, 8))
+  );
+  test_mug("Balanced tree depth 3", tree3);
+
+  /* Left-heavy tree */
+  u3_noun left_heavy = u3nc(u3nc(u3nc(1, 2), 3), 4);
+  test_mug("Left-heavy tree", left_heavy);
+
+  /* Right-heavy tree */
+  u3_noun right_heavy = u3nc(1, u3nc(2, u3nc(3, 4)));
+  test_mug("Right-heavy tree", right_heavy);
+  printf("\n");
+
+  printf("LISTS - Null-terminated lists:\n");
+  printf("─────────────────────────────────────\n");
+  /* [1 2 3 ~] = [1 [2 [3 0]]] */
+  u3_noun list_123 = u3nc(1, u3nc(2, u3nc(3, 0)));
+  test_mug("[1 2 3 ~]", list_123);
+
+  /* [0 ~] = [0 0] */
+  test_mug("[0 ~]", u3nc(0, 0));
+
+  /* [42 ~] = [42 0] */
+  test_mug("[42 ~]", u3nc(42, 0));
+
+  /* Empty list: ~ = 0 */
+  test_mug("~ (null/empty list)", 0);
+
+  /* Longer list [1 2 3 4 5 ~] */
+  u3_noun list_12345 = u3nc(1, u3nc(2, u3nc(3, u3nc(4, u3nc(5, 0)))));
+  test_mug("[1 2 3 4 5 ~]", list_12345);
+  printf("\n");
+
+  printf("SPECIAL - Urbit constants:\n");
+  printf("─────────────────────────────────────\n");
+  /* YES and NO in Urbit */
+  test_mug("YES (0)", 0);
+  test_mug("NO (1)", 1);
+
+  /* Common Urbit atoms */
+  test_mug("~zod (0)", 0);
+  test_mug("~nec (1)", 1);
+  test_mug("~bud (2)", 2);
+
+  /* Cord atoms (text) */
+  test_mug("'hello'", _cord("hello"));
+  test_mug("'world'", _cord("world"));
+  test_mug("'urbit'", _cord("urbit"));
+  test_mug("'nock'", _cord("nock"));
+  printf("\n");
+
+  printf("EDGE CASES - Repeated values:\n");
+  printf("─────────────────────────────────────\n");
+  /* Same cell computed multiple times should have same mug */
+  u3_noun pair = u3nc(123, 456);
+  test_mug("[123 456] - first computation", pair);
+  test_mug("[123 456] - second computation", u3nc(123, 456));
+
+  /* Nested identical cells */
+  u3_noun nested = u3nc(pair, pair);
+  test_mug("[[123 456] [123 456]]", nested);
+  printf("\n");
+
+  printf("═══════════════════════════════════════\n");
+  printf("Test complete! Compare with OCaml.\n");
+  printf("═══════════════════════════════════════\n\n");
+
+  return 0;
+}