many things, cosmo build!

.gitignore

@@ -12,4 +12,7 @@
 
 Player/player
 Player/player.wasm
+Player/Cosmo
 Master/master
+
+Player/Out/Cosmo/player*

Common/Std/any.c

@@ -0,0 +1,138 @@
+/*
+  Generic std parts, building on top of platform specifics.
+*/
+
+#include "../fix.h"
+#include "../std.h"
+#include "../str.h"
+#include <stdarg.h>
+#include <stdint.h>
+
+static uint8_t com_std_any_format_u32(uint32_t v, char buf[10]) {
+  uint8_t c = 0;
+  if (v == 0) {
+    buf[10 - (++c)] = '0';
+  } else {
+    while (v > 0) {
+      buf[10 - (++c)] = '0' + (v % 10);
+      v /= 10;
+    }
+  }
+  return c;
+}
+
+static uint8_t com_std_any_format_i32(int32_t v, char buf[11]) {
+  bool neg = v < 0;
+  uint8_t c = 0;
+  if (v == 0) {
+    buf[11 - (++c)] = '0';
+  } else {
+    while (v > 0) {
+      buf[11 - (++c)] = '0' + (v % 10);
+      v /= 10;
+    }
+  }
+  if (neg)
+    buf[11 - (++c)] = '-';
+  return c;
+}
+
+/* Just assume 5 points of precision. */
+static uint8_t com_std_any_format_f64(double v, char buf[11 + 1 + 5]) {
+  bool neg = v < 0;
+  uint8_t c = 0;
+  double frac_part = v - (double)(int32_t)v;
+  for (int i = 0; i < 5; i++) {
+    frac_part *= 10.0;
+    uint8_t digit = (unsigned int)frac_part;
+    buf[17 - (++c)] = digit + '0';
+    frac_part -= digit;
+  }
+  buf[17 - (++c)] = '.';
+  c += com_std_any_format_i32((int32_t)v, &buf[c]);
+  return c;
+}
+
+/* We can make it buffered, but who cares, it's only for testing. */
+void com_std_print_(int32_t start, ...) {
+  va_list args;
+  va_start(args, start);
+
+  bool afterstring = false;
+  char type = start;
+  do {
+
+    /* Put space if not proceeding string. */
+    if (!afterstring && type != 'c')
+      com_std_write(COM_STD_OUTFD, " ", 1);
+
+    afterstring = false;
+
+    switch (type) {
+
+    /* uint32_t case. */
+    case 'u': {
+      uint32_t const v = va_arg(args, uint32_t);
+      char buf[10];
+      uint8_t c = com_std_any_format_u32(v, buf);
+      com_std_write(COM_STD_OUTFD, &buf[sizeof buf - c], c);
+      break;
+    }
+
+    /* null terminated string case. */
+    case 'c': {
+      char const *const v = va_arg(args, char *);
+      com_std_write(COM_STD_OUTFD, v, com_str_ext_len(v));
+      afterstring = true;
+      break;
+    }
+
+    /* TODO: FIX THIS */
+    /* com_fix case. */
+    case 'f': {
+      com_fix_t const v = va_arg(args, com_fix_t);
+      char buf[11];
+
+      int32_t const int_part = com_fix_int(v);
+      uint8_t c = com_std_any_format_i32(int_part, buf);
+      com_std_write(COM_STD_OUTFD, &buf[sizeof buf - c], c);
+
+      uint32_t const decimal_val = (com_fix_frac(v) * 10000) >> 16;
+      uint8_t cd = com_std_any_format_u32(decimal_val, buf);
+      buf[sizeof buf - (++cd)] = '.';
+      com_std_write(COM_STD_OUTFD, &buf[sizeof buf - cd], cd);
+
+      break;
+    }
+
+    /* double floating point case. */
+    case 'd': {
+      double const v = va_arg(args, double);
+      char buf[17];
+
+      uint8_t c = com_std_any_format_f64(v, buf);
+      com_std_write(COM_STD_OUTFD, &buf[sizeof buf - c], c);
+    }
+
+    /* End of things. */
+    case '.':
+      break;
+
+    default:
+      com_std_assert(false && "Unhandled format type");
+    }
+
+    type = va_arg(args, int32_t);
+
+  } while (type != '.');
+
+  com_std_write(COM_STD_OUTFD, "\n", 1);
+
+  va_end(args);
+}
+
+void com_std_print_str(char const *str, uint32_t len, bool newline) {
+  com_std_write(COM_STD_OUTFD, str, len);
+  if (newline)
+    com_std_write(COM_STD_OUTFD, "\n", 1);
+}

Common/Std/libc.c

@@ -1,7 +1,9 @@
 #include "../std.h"
-#include "../str.h"
+#include "any.c"
 #include <assert.h>
+#include <poll.h>
 #include <stdarg.h>
+#include <stdint.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <sys/socket.h>
@@ -11,13 +13,7 @@
 void *com_std_alloc(void *ptr, size_t size) {
   return size != 0 ? realloc(ptr, size) : (free(ptr), NULL);
 }
-void com_std_assert(bool holds) { assert(holds); }
+
-void com_std_printf(char const *fmt, ...) {
-  va_list args;
-  va_start(args, fmt);
-  vprintf(fmt, args);
-  va_end(args);
-}
 double com_std_sqrt(double v) { return sqrt(v); }
 double com_std_fabs(double v) { return fabs(v); }
 double com_std_sin(double v) { return sin(v); }
@@ -43,6 +39,15 @@ int32_t com_std_read(com_std_fd fd, void *data, uint32_t upto) {
   return fread(data, 1, upto, fd);
 }
 
+bool com_std_has_data(com_std_fd fd) {
+  struct pollfd pfd = {.fd = (intptr_t)fd, .events = POLLIN};
+  int ret = poll(&pfd, 1, 0);
+  if (ret > 0) {
+    return pfd.revents & POLLIN;
+  } else
+    return false;
+}
+
 bool com_std_sfwd(com_std_fd fd, uint32_t bytes) {
   com_std_assert(fd != NULL && bytes != 0);
   return fseek(fd, bytes, SEEK_CUR) == 0;

Common/Std/linux-amd64.c

@@ -6,6 +6,7 @@
 
 #include "../std.h"
 #include <fcntl.h>
+#include <poll.h>
 #include <stdint.h>
 #include <sys/socket.h>
 #include <sys/un.h>
@@ -44,6 +45,23 @@ inline int32_t com_std_read(com_std_fd fd, void *data, uint32_t upto) {
   return ret;
 }
 
+inline bool com_std_has_data(com_std_fd fd) {
+  struct pollfd pfd = {
+      .fd = (intptr_t)fd,
+      .events = POLLIN,
+  };
+
+  long ret;
+  __asm__ volatile("syscall"
+                   : "=a"(ret)
+                   : "a"(7), "D"(&pfd), "S"(1), "d"(0)
+                   : "rcx", "r11", "memory");
+  if (ret > 0) {
+    return pfd.revents & POLLIN;
+  } else
+    return false;
+}
+
 inline bool com_std_sfwd(com_std_fd fd, uint32_t bytes) {
   long ret;
   __asm__ volatile("syscall"

Common/Std/linux-x86.c

@@ -6,6 +6,7 @@
 
 #include "../std.h"
 #include <fcntl.h>
+#include <poll.h>
 #include <stdint.h>
 #include <sys/socket.h>
 #include <sys/un.h>
@@ -44,6 +45,23 @@ inline int32_t com_std_read(com_std_fd fd, void *data, uint32_t upto) {
   return ret;
 }
 
+inline bool com_std_has_data(com_std_fd fd) {
+  struct pollfd pfd = {
+      .fd = (intptr_t)fd,
+      .events = POLLIN,
+  };
+
+  long ret;
+  __asm__ volatile("int $0x80"
+                   : "=a"(ret)
+                   : "a"(168), "b"(&pfd), "c"(1), "d"(0)
+                   : "memory", "cc");
+  if (ret > 0) {
+    return pfd.revents & POLLIN;
+  } else
+    return false;
+}
+
 inline bool com_std_sfwd(com_std_fd fd, uint32_t bytes) {
   long ret;
   __asm__ volatile("int $0x80"

Common/Std/linux.c

@@ -2,41 +2,13 @@
   Generic linux std parts.
 */
 
-#include "../def.h"
+#include "any.c"
+#include "nolibc.c"
+
 #include "../std.h"
-#include "../str.h"
-#include <stdarg.h>
 #include <stdint.h>
 
 void *com_std_alloc(void *ptr, size_t size) { return NULL; }
-void com_std_assert(bool holds) {
-  /* TODO: Write line info to stderr as well. */
-  if (!holds)
-    com_std_exit(-1);
-}
-
-/* TODO: Figure out a way to make it easy to deal with. */
-void com_std_printf(char const *fmt, ...) {
-  com_std_write((void *)1, fmt, com_str_ext_len(fmt));
-}
-
-void com_std_print_str(char const *str, uint32_t len, bool newline) {
-  com_std_write((void *)1, str, len);
-  if (newline)
-    com_std_write((void *)1, "\n", 1);
-}
 
 double com_std_sqrt(double v) { return __builtin_sqrt(v); }
 double com_std_fabs(double v) { return __builtin_fabs(v); }
-#if COM_DEF_COMPILER == COM_DEF_COMPILER_CLANG
-double com_std_sin(double v) { return __builtin_sin(v); }
-double com_std_cos(double v) { return __builtin_cos(v); }
-#else
-/* TODO: Use fsin/fcos, no matter how slow. */
-double com_std_sin(double v) { return 0; }
-double com_std_cos(double v) { return 0; }
-#endif
-
-// void memcpy(void *__restrict dest, const void *__restrict src, size_t n) {
-//   __builtin_memcpy(dest, src, n);
-// };

Common/Std/nolibc.c

@@ -0,0 +1,46 @@
+/*
+  Generic no libc std parts.
+*/
+
+#include <stddef.h>
+
+/* Compiler depends on them with low optimization level. */
+void *memcpy(void *__restrict dest, const void *__restrict src, size_t n) {
+  for (size_t i = 0; i < n; ++i)
+    ((char *)dest)[i] = ((char *)src)[i];
+  return dest;
+};
+
+void *memset(void *__restrict dest, int s, size_t n) {
+  for (size_t i = 0; i < n; ++i)
+    ((char *)dest)[i] = s;
+  return dest;
+};
+
+size_t strlen(char const *s) {
+  char const *p = s;
+  for (; *p != 0; ++p) {
+  }
+  return p - s;
+}
+
+double com_std_sin(double v) {
+  double result;
+  __asm__ __volatile__("fldl %1\n\t"
+                       "fsin\n\t"
+                       "fstpl %0"
+                       : "=g"(result)
+                       : "m"(v)
+                       : "st", "st(1)");
+  return result;
+}
+
+double com_std_cos(double v) {
+  double result;
+  __asm__ __volatile__("fldl %1\n\t"
+                       "fcos\n\t"
+                       "fstpl %0"
+                       : "=g"(result)
+                       : "m"(v));
+  return result;
+}

Common/Std/stub.c

@@ -1,11 +1,11 @@
 #include "../std.h"
 
+#include "any.c"
+
 #include <stdbool.h>
 #include <stddef.h>
 
 void *com_std_alloc(void *ptr, size_t size) { return NULL; }
-void com_std_assert(bool holds) {}
-void com_std_printf(char const *fmt, ...) {}
 
 double com_std_sqrt(double v) { return __builtin_sqrt(v); }
 double com_std_fabs(double v) { return __builtin_fabs(v); }
@@ -35,7 +35,3 @@ void com_std_exit(uint32_t status) {
   __builtin_unreachable();
 }
 void com_std_close(com_std_fd fd) { return; }
-
-// void memcpy(void *__restrict dest, const void *__restrict src, size_t n) {
-//   __builtin_memcpy(dest, src, n);
-// };

Common/Timer/posix.c

@@ -13,5 +13,6 @@ uint64_t com_timer_count_ns(void) {
 void com_timer_profile(uint64_t starttime, const char *what) {
   uint64_t const endtime = com_timer_count_ns();
   uint64_t const diff = endtime - starttime;
-  com_std_printf("profile \"%s\" took %fms\n", what, (double)diff / 1000000);
+  com_std_print('c', "Profile ", 'c', what, 'c', " took ", 'd',
+                (double)diff / 1000000, 'c', "ms");
 }

Common/cfg.h

@@ -11,4 +11,12 @@
 #define COM_CFG_WIDTH 640
 #define COM_CFG_HEIGHT 480
 
+#if COM_DEF_TARGET == COM_DEF_TARGET_WASM
+#define COM_CFG_COLORS 4
+#else
+#define COM_CFG_COLORS 3
+#endif
+
+#define COM_CFG_FBSIZE (COM_CFG_WIDTH * COM_CFG_HEIGHT * COM_CFG_COLORS)
+
 #endif

Common/def.h

@@ -6,6 +6,9 @@
 #ifndef COM_DEF_H
 #define COM_DEF_H
 
+#define COM_DEF_STRLIT_(x) #x
+#define COM_DEF_STRLIT(x) COM_DEF_STRLIT_(x)
+
 #define COM_DEF_COMPILER_CLANG 1
 #define COM_DEF_COMPILER_GCC 2
 
@@ -17,6 +20,7 @@
 
 #define COM_DEF_TARGET_LINUX 1
 #define COM_DEF_TARGET_WASM 2
+#define COM_DEF_TARGET_COSMO 3
 
 #if defined(__wasm__)
 #define COM_DEF_TARGET COM_DEF_TARGET_WASM
@@ -24,6 +28,9 @@
 #elif defined(__linux__)
 #define COM_DEF_TARGET COM_DEF_TARGET_LINUX
 #define com_def_export_sym(m_s) __attribute__((visibility("default")))
+#elif defined(__COSMOPOLITAN__)
+#define COM_DEF_TARGET COM_DEF_TARGET_COSMO
+#define com_def_export_sym(m_s) __attribute__((visibility("default")))
 #else
 #error Unknown target.
 #endif
@@ -31,6 +38,7 @@
 #define COM_DEF_CPU_AMD64 1
 #define COM_DEF_CPU_X86 2
 #define COM_DEF_CPU_WASM32 3
+#define COM_DEF_CPU_AARCH64 4
 
 #if defined(__wasm__)
 #define COM_DEF_CPU COM_DEF_CPU_WASM32
@@ -44,6 +52,10 @@
 #define COM_DEF_CPU COM_DEF_CPU_X86
 #define COM_DEF_ALIGNMENT 16
 #define COM_DEF_CPU_LITEND 1
+#elif defined(__aarch64__)
+#define COM_DEF_CPU COM_DEF_CPU_AARCH64
+#define COM_DEF_ALIGNMENT 16
+#define COM_DEF_CPU_LITEND 1
 #else
 #error Unknown cpu.
 #endif

Common/fix.c

@@ -71,20 +71,6 @@ com_fix_t com_fix_sqrt(com_fix_t a) {
                          : (int32_t)(x >> (-final_shift));
 }
 
-void com_fix_print(com_fix_t a) {
-  if (a < 0) {
-    com_std_printf("-");
-    a = -a;
-  }
-  int32_t int_part = a >> 16;
-  int32_t frac_part = a & 0xFFFF;
-
-  // Convert fractional 16-bit part to a decimal value (up to 4 decimal places)
-  uint32_t decimal_val = (frac_part * 10000) >> 16;
-
-  com_std_printf("%d.%04u\n", int_part, decimal_val);
-}
-
 void com_fix_run_bench(void) {
   uint64_t start = com_timer_count_ns();
   for (int i = 5000000; i--;) {
@@ -118,77 +104,3 @@ void com_fix_run_bench(void) {
   }
   com_timer_profile(start, "fixed: sincos");
 }
-
-void com_fix_run_tests(void) {
-  {
-    double max_sqrt_deviation = 0.0f;
-    com_fix_t sqrt_accumulator = 0;
-    com_fix_t sqrt_test_step = COM_FIX_FRACUNIT >> 2; /* Quarter step */
-    for (int i = 0; i < COM_FIX_FRACUNIT << 1; ++i) {
-      com_fix_t sqrt = com_fix_sqrt(sqrt_accumulator);
-      double const deviation =
-          com_std_sqrt(com_fix_as_float(sqrt_accumulator)) -
-          com_fix_as_float(sqrt);
-      if (deviation > max_sqrt_deviation)
-        max_sqrt_deviation = deviation;
-      sqrt_accumulator += sqrt_test_step;
-    }
-    com_std_printf("max sqrt deviation: %f\n", max_sqrt_deviation);
-  }
-
-  {
-    double max_sin_deviation = 0.0f;
-    com_fix_t sin_accumulator = -COM_FIX_PI * 4;
-    com_fix_t sin_test_step = (COM_FIX_PI << 1) >> 10;
-    for (int i = 0; i < 1024 * 32; ++i) {
-      com_fix_t sin = com_fix_sin(sin_accumulator);
-      double const deviation =
-          com_std_fabs(com_std_sin(com_fix_as_float(sin_accumulator)) -
-                       com_fix_as_float(sin));
-      if (deviation > max_sin_deviation)
-        max_sin_deviation = deviation;
-      sin_accumulator += sin_test_step;
-    }
-    com_std_printf("max sin deviation: %f\n", max_sin_deviation);
-  }
-
-  {
-    double max_cos_deviation = 0.0f;
-    com_fix_t cos_accumulator = -COM_FIX_PI * 4;
-    com_fix_t cos_test_step = (COM_FIX_PI << 1) >> 10;
-    for (int i = 0; i < 1024 * 32; ++i) {
-      com_fix_t cos = com_fix_cos(cos_accumulator);
-      double const deviation =
-          com_std_fabs(com_std_cos(com_fix_as_float(cos_accumulator)) -
-                       com_fix_as_float(cos));
-      if (deviation > max_cos_deviation)
-        max_cos_deviation = deviation;
-      cos_accumulator += cos_test_step;
-    }
-    com_std_printf("max cos deviation: %f\n", max_cos_deviation);
-  }
-
-  {
-    double max_sin_deviation = 0.0f;
-    double max_cos_deviation = 0.0f;
-    com_fix_t sincos_accumulator = -COM_FIX_PI * 4;
-    com_fix_t sincos_test_step = (COM_FIX_PI << 1) >> 10;
-    for (int i = 0; i < 1024 * 32; ++i) {
-      com_fix_t sin, cos;
-      com_fix_sincos(sincos_accumulator, &sin, &cos);
-      double const sin_deviation =
-          com_std_fabs(com_std_sin(com_fix_as_float(sincos_accumulator)) -
-                       com_fix_as_float(sin));
-      double const cos_deviation =
-          com_std_fabs(com_std_cos(com_fix_as_float(sincos_accumulator)) -
-                       com_fix_as_float(cos));
-      if (sin_deviation > max_sin_deviation)
-        max_sin_deviation = sin_deviation;
-      if (cos_deviation > max_cos_deviation)
-        max_cos_deviation = cos_deviation;
-      sincos_accumulator += sincos_test_step;
-    }
-    com_std_printf("max sincos deviations: %f %f\n", max_sin_deviation,
-                   max_cos_deviation);
-  }
-}

Common/fix.h

@@ -23,8 +23,6 @@ typedef int32_t com_fix_t;
 
 extern const uint16_t com_fix_sin_lut[128];
 
-void com_fix_print(com_fix_t a);
-
 void com_fix_run_tests(void);
 void com_fix_run_bench(void);
 
@@ -32,9 +30,9 @@ static inline com_fix_t com_fix_add(com_fix_t a, com_fix_t b) { return a + b; }
 
 static inline com_fix_t com_fix_sub(com_fix_t a, com_fix_t b) { return a - b; }
 
-static inline com_fix_t com_fix_int(com_fix_t a) {
+static inline int16_t com_fix_int(com_fix_t a) { return a >> COM_FIX_FRACBITS; }
-  return a >> COM_FIX_FRACBITS;
+
-}
+static inline uint16_t com_fix_frac(com_fix_t a) { return a & 0xFFFF; }
 
 static inline com_fix_t com_fix_mul(com_fix_t a, com_fix_t b) {
   return (com_fix_t)(((int64_t)a * (int64_t)b) >> COM_FIX_FRACBITS);

Common/gem.h

@@ -91,9 +91,9 @@ static inline void com_gem_draw_triangle(com_pnt_t v0, com_pnt_t v1,
       // If the point is inside or on all edges, draw the pixel
       // (Assumes Counter Clockwise vertex ordering)
       if (ws.a[0] >= 0 && ws.a[1] >= 0 && ws.a[2] >= 0) {
-        out[(y * COM_CFG_WIDTH + x) * 4 + 0] = 125;
+        out[(y * COM_CFG_WIDTH + x) * COM_CFG_COLORS + 0] = 125;
-        out[(y * COM_CFG_WIDTH + x) * 4 + 1] = 125;
+        out[(y * COM_CFG_WIDTH + x) * COM_CFG_COLORS + 1] = 125;
-        out[(y * COM_CFG_WIDTH + x) * 4 + 2] = 125;
+        out[(y * COM_CFG_WIDTH + x) * COM_CFG_COLORS + 2] = 125;
       }
     }
   }
@@ -147,9 +147,9 @@ com_gem_draw_triangle_textured_edge(com_pnt_t v0, com_pnt_t v1, com_pnt_t v2,
         // uint32_t bidx = com_zord_order(x, y);
         uint32_t bidx = y * COM_CFG_WIDTH + x;
 
-        out[bidx * 4 + 0] = tex[idx * 3 + 0];
+        out[bidx * COM_CFG_COLORS + 0] = tex[idx * 3 + 0];
-        out[bidx * 4 + 1] = tex[idx * 3 + 1];
+        out[bidx * COM_CFG_COLORS + 1] = tex[idx * 3 + 1];
-        out[bidx * 4 + 2] = tex[idx * 3 + 2];
+        out[bidx * COM_CFG_COLORS + 2] = tex[idx * 3 + 2];
       }
     }
   }
@@ -170,9 +170,9 @@ static inline void draw_span(int32_t x_start_fp, int32_t x_end_fp, int y,
   for (int x = x1; x < x2; x++) {
     uint32_t idx = 0 * 32 + 0;
     uint32_t bidx = y * COM_CFG_WIDTH + x;
-    out[bidx * 4 + 0] = tex[idx * 3 + 0];
+    out[bidx * COM_CFG_COLORS + 0] = tex[idx * 3 + 0];
-    out[bidx * 4 + 1] = tex[idx * 3 + 1];
+    out[bidx * COM_CFG_COLORS + 1] = tex[idx * 3 + 1];
-    out[bidx * 4 + 2] = tex[idx * 3 + 2];
+    out[bidx * COM_CFG_COLORS + 2] = tex[idx * 3 + 2];
   }
 }
 

Common/pnt.h

@@ -111,9 +111,4 @@ com_pnt_edge_weight_calc_precomp(com_pnt_t v0, com_pnt_t v1, com_pnt_t v2,
                                precomp[2].a[1] * (p.a[0] - v0.a[0])}};
 }
 
-static inline void com_pnt_print(com_pnt_t a) {
-  com_fix_print(a.a[0]);
-  com_fix_print(a.a[1]);
-}
-
 #endif

Common/std.h

@@ -10,6 +10,16 @@
 #include <stddef.h>
 #include <stdint.h>
 
+#if COM_DEF_TARGET == COM_DEF_TARGET_WASM
+#define COM_STD_INFD 0
+#define COM_STD_OUTFD 1
+#else
+/* TODO: Make it more graceful, probably with static func. */
+#include <unistd.h>
+#define COM_STD_INFD (fdopen(STDIN_FILENO, "r"))
+#define COM_STD_OUTFD (fdopen(STDOUT_FILENO, "w"))
+#endif
+
 /* Generic system interface, linux/bsd derived. */
 typedef void *com_std_fd;
 #define COM_STD_OPEN_READ (1 << 0)
@@ -17,6 +27,7 @@ typedef void *com_std_fd;
 com_std_fd com_std_open(char const *path, uint32_t flags);
 bool com_std_write(com_std_fd fd, void const *data, uint32_t bytes);
 int32_t com_std_read(com_std_fd fd, void *data, uint32_t upto);
+bool com_std_has_data(com_std_fd);
 bool com_std_sfwd(com_std_fd fd, uint32_t bytes);
 com_std_fd com_std_socket(bool isnetwork);
 bool com_std_connect(com_std_fd sock, bool isnetwork, char const *address);
@@ -31,9 +42,44 @@ void *com_std_alloc(void *ptr, size_t size);
 /* Note: Functions listed below are only available for dev builds. */
 /* Trigonometry functions are for reference testing of our own fixed point
  * implementations. Do not use otherwise. */
-void com_std_assert(bool holds);
+#define com_std_assert(p_h)                                                    \
-void com_std_printf(char const *fmt, ...);
+  do {                                                                         \
+    if (!(p_h)) {                                                              \
+      const char msg[] =                                                       \
+          "Assertion failed at "__FILE__                                       \
+          ":" COM_DEF_STRLIT(__LINE__) ": " COM_DEF_STRLIT(p_h);               \
+      com_std_print_str(msg, sizeof msg, true);                                \
+      com_std_exit(-1);                                                        \
+    }                                                                          \
+  } while (0)
+
+/*
+    Basic printing utility that doesn't utilize format string.
+    Instead a vector of arguments is passed and concatenated, based on type.
+    Newline is implied at the end.
+
+    Example: com_std_print("My unit32_t: ", 'u', value, 's', " is here");
+
+    Format specifiers:
+    'u' = uint32_t
+    'i' = int32_t
+    'd' = float (upcast to double with vargs)
+    'c' = null terminated C string
+
+    Our own types:
+    'f' = com_fix value
+    'p' = com_pnt pointer
+    'v' = com_vec pointer
+    's' = pointer to com_str structure
+    'b' = pointer to com_buf structure
+
+    '.' is placed at the end implicitly with macro,
+        it is needed to know when to end.
+*/
+void com_std_print_(int32_t start, ...);
+#define com_std_print(...) com_std_print_(__VA_ARGS__, '.')
 void com_std_print_str(char const *str, uint32_t len, bool newline);
+
 double com_std_sqrt(double v);
 double com_std_fabs(double v);
 double com_std_sin(double v);

Common/test.h

@@ -0,0 +1,121 @@
+/*
+    Centralized test suite.
+*/
+
+#include "bits.h"
+#include "fix.h"
+#include "std.h"
+#include <stdbool.h>
+
+#define TEST(p_e)                                                              \
+  do {                                                                         \
+    if (!(p_e)) {                                                              \
+      com_std_print('c', COM_DEF_STRLIT(p_e) " failed!");                      \
+      anyfail = true;                                                          \
+    }                                                                          \
+  } while (0)
+
+static bool com_test_fix(void) {
+  bool anyfail = false;
+  {
+    double max_sqrt_deviation = 0.0f;
+    com_fix_t sqrt_accumulator = 0;
+    com_fix_t sqrt_test_step = COM_FIX_FRACUNIT >> 2; /* Quarter step */
+    for (int i = 0; i < COM_FIX_FRACUNIT << 1; ++i) {
+      com_fix_t sqrt = com_fix_sqrt(sqrt_accumulator);
+      double const deviation =
+          com_std_sqrt(com_fix_as_float(sqrt_accumulator)) -
+          com_fix_as_float(sqrt);
+      if (deviation > max_sqrt_deviation)
+        max_sqrt_deviation = deviation;
+      sqrt_accumulator += sqrt_test_step;
+    }
+    com_std_print('c', "Max sqrt deviation: ", 'd', max_sqrt_deviation);
+  }
+
+  {
+    double max_sin_deviation = 0.0f;
+    com_fix_t sin_accumulator = -COM_FIX_PI * 4;
+    com_fix_t sin_test_step = (COM_FIX_PI << 1) >> 10;
+    for (int i = 0; i < 1024 * 32; ++i) {
+      com_fix_t sin = com_fix_sin(sin_accumulator);
+      double const deviation =
+          com_std_fabs(com_std_sin(com_fix_as_float(sin_accumulator)) -
+                       com_fix_as_float(sin));
+      if (deviation > max_sin_deviation)
+        max_sin_deviation = deviation;
+      sin_accumulator += sin_test_step;
+    }
+    com_std_print('c', "Max sin deviation: ", 'd', max_sin_deviation);
+  }
+
+  {
+    double max_cos_deviation = 0.0f;
+    com_fix_t cos_accumulator = -COM_FIX_PI * 4;
+    com_fix_t cos_test_step = (COM_FIX_PI << 1) >> 10;
+    for (int i = 0; i < 1024 * 32; ++i) {
+      com_fix_t cos = com_fix_cos(cos_accumulator);
+      double const deviation =
+          com_std_fabs(com_std_cos(com_fix_as_float(cos_accumulator)) -
+                       com_fix_as_float(cos));
+      if (deviation > max_cos_deviation)
+        max_cos_deviation = deviation;
+      cos_accumulator += cos_test_step;
+    }
+    com_std_print('c', "Max cos deviation: ", 'd', max_cos_deviation);
+  }
+
+  {
+    double max_sin_deviation = 0.0f;
+    double max_cos_deviation = 0.0f;
+    com_fix_t sincos_accumulator = -COM_FIX_PI * 4;
+    com_fix_t sincos_test_step = (COM_FIX_PI << 1) >> 10;
+    for (int i = 0; i < 1024 * 32; ++i) {
+      com_fix_t sin, cos;
+      com_fix_sincos(sincos_accumulator, &sin, &cos);
+      double const sin_deviation =
+          com_std_fabs(com_std_sin(com_fix_as_float(sincos_accumulator)) -
+                       com_fix_as_float(sin));
+      double const cos_deviation =
+          com_std_fabs(com_std_cos(com_fix_as_float(sincos_accumulator)) -
+                       com_fix_as_float(cos));
+      if (sin_deviation > max_sin_deviation)
+        max_sin_deviation = sin_deviation;
+      if (cos_deviation > max_cos_deviation)
+        max_cos_deviation = cos_deviation;
+      sincos_accumulator += sincos_test_step;
+    }
+    com_std_print('c', "Max sincos deviations: ", 'd', max_sin_deviation, 'd',
+                  max_cos_deviation);
+  }
+
+  return anyfail;
+}
+
+static bool com_test_bits(void) {
+  bool anyfail = false;
+
+  TEST(com_bits_nalign(0, 4) == 0);
+  TEST(com_bits_nalign(1, 4) == 4);
+  TEST(com_bits_nalign(2, 4) == 4);
+  TEST(com_bits_nalign(3, 4) == 4);
+  TEST(com_bits_nalign(4, 4) == 4);
+  TEST(com_bits_nalign(5, 4) == 8);
+
+  return anyfail;
+}
+
+static bool com_test_run(void) {
+  bool anyfail = false;
+  if (com_test_fix()) {
+    com_std_print('c', "com_fix tests failed!");
+    anyfail = true;
+  }
+  if (com_test_bits()) {
+    com_std_print('c', "com_bits tests failed!");
+    anyfail = true;
+  }
+  return !anyfail;
+}
+
+#undef TEST

Common/vec.h

@@ -93,10 +93,4 @@ static inline com_fix_t com_vec_max(com_vec_t a) {
   return com_fix_max(com_fix_max(a.a[0], a.a[1]), a.a[2]);
 }
 
-static inline void com_vec_print(com_vec_t a) {
-  com_fix_print(a.a[0]);
-  com_fix_print(a.a[1]);
-  com_fix_print(a.a[2]);
-}
-
 #endif

Player/Display/display.h

@@ -1,7 +1,8 @@
 #ifndef PLR_DISPLAY_H
 #define PLR_DISPLAY_H
 
-__attribute__((used)) extern int plr_display_x11_main(int argc, char *argv[]);
+__attribute__((used)) extern int plr_display_x11_main(int argc,
+                                                      char const *const *argv);
 __attribute__((used)) extern int plr_display_dos_main(void);
 
 #endif

Player/Display/x11.c

@@ -1,6 +1,3 @@
-#include <X11/X.h>
-#include <X11/Xlib.h>
-#include <X11/Xutil.h>
 #include <fcntl.h>
 #include <stdbool.h>
 #include <stdint.h>
@@ -11,16 +8,26 @@
 #include "../../Common/bits.h"
 #include "../../Common/buf.h"
 #include "../../Common/cfg.h"
+#include "../../Common/def.h"
 #include "../../Common/lzw.h"
 #include "../../Common/mat.h"
 #include "../../Common/std.h"
 #include "../../Common/str.h"
+#include "../../Common/test.h"
 #include "../draw.h"
 #include "display.h"
 
-/* https://hereket.com/posts/from-scratch-x11-windowing/ */
+#if COM_DEF_TARGET == COM_DEF_TARGET_COSMO
-/* https://gaultier.github.io/blog/write_a_video_game_from_scratch_like_1987.html
+#include <X.h>
- */
+#else
+#include <X11/X.h>
+#endif
+
+/*
+  https://hereket.com/posts/from-scratch-x11-windowing/
+  https://gaultier.github.io/blog/write_a_video_game_from_scratch_like_1987.html
+  https://xorg.freedesktop.org/archive/X11R7.7-RC1/doc/xproto/x11protocol.html
+*/
 
 static struct {
   uint32_t id;
@@ -94,7 +101,7 @@ static struct plr_x11_auth_entry {
   result.success = true;
   return result;
 BREAK:
-  com_std_printf("Failure to initialize X11\n");
+  com_std_print('c', "Failure to initialize X11");
   return result;
 }
 
@@ -108,8 +115,8 @@ static bool plr_x11_handshake(com_std_fd xfd,
     uint16_t auth_len;
     uint16_t auth_data_len;
     uint16_t pad1;
-  } request = {
+  } __attribute__((packed)) request = {
-      .endianness = 'l',
+      .endianness = COM_DEF_CPU_LITEND ? 'l' : 'B',
       .major_version = 11,
       .auth_len = sizeof(AUTH_ENTRY_MAGIC_COOKIE) - 1, // IS -1 NEEDED?
       .auth_data_len = auth_entry.auth_data->len,
@@ -131,7 +138,7 @@ static bool plr_x11_handshake(com_std_fd xfd,
     uint16_t major_version;
     uint16_t minor_version;
     uint16_t length;
-  } response;
+  } __attribute__((packed)) response;
 
   if (!com_std_read(xfd, &response, sizeof response))
     return false;
@@ -154,7 +161,7 @@ static bool plr_x11_handshake(com_std_fd xfd,
     uint8_t min_keycode;
     uint8_t max_keycode;
     uint32_t pad2;
-  } dyn_response;
+  } __attribute__((packed)) dyn_response;
 
   if (!com_std_read(xfd, &dyn_response, sizeof dyn_response))
     return false;
@@ -177,6 +184,54 @@ static bool plr_x11_handshake(com_std_fd xfd,
   return true;
 }
 
+/* TODO: Instead have a poll_events function instead. */
+static inline bool plr_x11_poll_errors(com_std_fd xfd) {
+  union {
+    struct {
+      uint8_t type;
+      uint8_t errcode;
+      uint16_t sequenceid;
+      uint32_t badval;
+      uint16_t minorop;
+      uint32_t majorop;
+      char pad[18];
+    } __attribute__((packed)) error;
+    struct {
+      uint8_t type;
+      uint8_t data;
+      uint16_t sequenceid;
+      uint32_t length;
+      char pad[24];
+    } __attribute__((packed)) reply;
+  } response;
+
+  while (com_std_has_data(xfd)) {
+    if (!com_std_read(xfd, &response, sizeof response))
+      return false;
+
+    /* Exhaust variable length data of reply. */
+    if (response.reply.type == 1) {
+      char buf[1024];
+      com_std_assert(response.reply.length * 4 <= sizeof buf);
+      if (!com_std_read(xfd, &buf, response.reply.length * 4))
+        return false;
+    }
+
+    if (response.reply.type == 0 && response.error.errcode != 0) {
+      com_std_print('c', "X11 error code=", 'u', response.error.errcode, 'c',
+                    " for seqid=", 'u', response.error.sequenceid);
+
+      /* This means X11 closed the connection. */
+      // if (response.error.errcode == 0)
+      //   return false;
+    }
+
+    /* TODO: Save polled events, as they might be used later. */
+  };
+
+  return true;
+}
+
 static inline uint32_t plr_x11_next_id(void) {
   static uint32_t cur_id = 0;
   cur_id = 1 + ((plr_x11_connection.resource_id_mask & (cur_id)) |
@@ -184,29 +239,35 @@ static inline uint32_t plr_x11_next_id(void) {
   return cur_id;
 }
 
+/* Used to number requests. */
+static inline uint32_t plr_x11_next_seqid(void) {
+  static uint32_t cur_id = 1;
+  return cur_id++;
+}
+
 static uint32_t plr_x11_create_graphical_context(com_std_fd xfd) {
   uint32_t const gc_id = plr_x11_next_id();
+  uint32_t const seq = plr_x11_next_seqid();
   struct {
     uint8_t opcode;
     uint8_t pad1;
-    uint16_t length;
+    uint16_t request_length;
     uint32_t id;
     uint32_t drawable;
     uint32_t bitmask;
     uint32_t value1;
-  } request = {
+  } __attribute__((packed)) request = {
       .opcode = 55,
-      .length = 5,
+      .request_length = sizeof request / 4,
       .id = gc_id,
       .drawable = plr_x11_root_screen.id,
       .bitmask = 8, /* TODO: Figure out. */
-      .value1 = 0x0000ff00,
+      .value1 = 0x00000000,
   };
 
-  if (com_std_write(xfd, &request, sizeof request))
+  if (!com_std_write(xfd, &request, sizeof request))
-    return gc_id;
-  else
     return 0;
+  return gc_id;
 }
 
 static uint32_t plr_x11_create_window(com_std_fd xfd) {
@@ -220,6 +281,7 @@ static uint32_t plr_x11_create_window(com_std_fd xfd) {
 #define EVENT_FLAG_BUTTON_RELEASE 0x8
 
   uint32_t const window_id = plr_x11_next_id();
+  uint32_t const seq = plr_x11_next_seqid();
   struct {
     uint8_t opcode;
     uint8_t depth;
@@ -236,10 +298,10 @@ static uint32_t plr_x11_create_window(com_std_fd xfd) {
     uint32_t bitmask;
     uint32_t value1;
     uint32_t value2;
-  } request = {
+  } __attribute__((packed)) request = {
       .opcode = 1,
-      .depth = 32,
+      .depth = COM_CFG_COLORS * 8,
-      .request_length = 8 + 2,
+      .request_length = sizeof request / 4,
       .window_id = window_id,
       .parent_id = plr_x11_root_screen.id,
       .width = COM_CFG_WIDTH,
@@ -248,16 +310,15 @@ static uint32_t plr_x11_create_window(com_std_fd xfd) {
       .class = 1,
       .root_visual_id = plr_x11_root_screen.root_visual_id,
       .bitmask = FLAG_WIN_BG_PIXEL | FLAG_WIN_EVENT,
-      .value1 = 0x00ffff00,
+      .value1 = 0x00000000,
       .value2 = EVENT_FLAG_EXPOSURE | EVENT_FLAG_BUTTON_RELEASE |
                 EVENT_FLAG_BUTTON_PRESS | EVENT_FLAG_KEY_PRESS |
                 EVENT_FLAG_KEY_RELEASE,
   };
 
-  if (com_std_write(xfd, &request, sizeof request))
+  if (!com_std_write(xfd, &request, sizeof request))
-    return window_id;
-  else
     return 0;
+  return window_id;
 }
 
 static bool plr_x11_show_window(com_std_fd xfd, uint32_t window_id) {
@@ -266,9 +327,9 @@ static bool plr_x11_show_window(com_std_fd xfd, uint32_t window_id) {
     uint8_t pad1;
     uint16_t request_length;
     uint32_t window_id;
-  } request = {
+  } __attribute__((packed)) request = {
       .opcode = 8,
-      .request_length = 2,
+      .request_length = sizeof request / 4,
       .window_id = window_id,
   };
   return com_std_write(xfd, &request, sizeof request);
@@ -276,36 +337,35 @@ static bool plr_x11_show_window(com_std_fd xfd, uint32_t window_id) {
 
 static uint32_t plr_x11_create_pixmap(com_std_fd xfd, uint32_t window_id) {
   uint32_t const pixmap_id = plr_x11_next_id();
+  uint32_t const seq = plr_x11_next_seqid();
   struct {
     uint8_t opcode;
     uint8_t depth;
-    uint16_t length;
+    uint16_t request_length;
     uint32_t pixmap_id;
     uint32_t drawable;
     uint16_t width;
     uint16_t height;
-  } request = {
+  } __attribute__((packed)) request = {
       .opcode = 53,
-      .depth = 32,
+      .depth = COM_CFG_COLORS * 8,
-      .length = 4,
+      .request_length = sizeof request / 4,
       .pixmap_id = pixmap_id,
       .drawable = window_id,
       .width = COM_CFG_WIDTH,
       .height = COM_CFG_HEIGHT,
   };
 
-  if (com_std_write(xfd, &request, sizeof request))
+  if (!com_std_write(xfd, &request, sizeof request))
-    return pixmap_id;
-  else
     return 0;
+  return pixmap_id;
 }
 
-static bool
+static bool plr_x11_put_framebuffer(com_std_fd xfd, uint32_t gc_id,
-plr_x11_put_framebuffer(com_std_fd xfd, uint32_t gc_id, uint32_t window_id,
                                     uint32_t pixmap_id,
-                        uint32_t pixels[COM_CFG_WIDTH * COM_CFG_HEIGHT]) {
+                                    uint8_t pixels[COM_CFG_FBSIZE]) {
-  uint32_t const padded =
+  uint32_t const seq = plr_x11_next_seqid();
-      com_bits_nalign(4 * COM_CFG_WIDTH * COM_CFG_HEIGHT, 4);
+  uint32_t const padded = com_bits_nalign(COM_CFG_FBSIZE, 4);
 
   struct {
     uint8_t opcode;
@@ -320,35 +380,84 @@ plr_x11_put_framebuffer(com_std_fd xfd, uint32_t gc_id, uint32_t window_id,
     uint8_t left_pad;
     uint8_t depth;
     uint16_t pad1;
-  } request = {
+  } __attribute__((packed)) request = {
       .opcode = 72,
       .format = 2, // ZPixmap
-      .request_length = 6 + padded / 4,
+      .request_length = sizeof request / 4 + padded / 4,
       .drawable_id = pixmap_id,
       .gc_id = gc_id,
       .width = COM_CFG_WIDTH,
       .height = COM_CFG_HEIGHT,
       .dst_x = 0,
       .dst_y = 0,
-      .depth = 32,
+      .depth = COM_CFG_COLORS * 8,
   };
 
   if (!com_std_write(xfd, &request, sizeof request))
     return false;
-  if (!com_std_write(xfd, pixels, 4 * COM_CFG_WIDTH * COM_CFG_HEIGHT))
+  if (!com_std_write(xfd, pixels, COM_CFG_FBSIZE))
     return false;
   /* Use request memory for pad, it doesn't matter what we write. */
-  uint32_t const padfinal = padded - (4 * COM_CFG_WIDTH * COM_CFG_HEIGHT);
+  uint32_t const padfinal = padded - COM_CFG_FBSIZE;
-  if (padfinal && !com_std_write(xfd, &request, padfinal))
+  if (padfinal && !com_std_write(xfd, &pixels, padfinal))
     return false;
+  return true;
+}
+
+static bool plr_x11_establish_connection(com_std_fd xfd, int argc,
+                                         char const *const *argv) {
+  com_buf_make_on_stack(char, xauthbuf, 1024, 0);
+
+  /* First we need to authenticate X11 connection, which requires some funk.
+   */
+  /* We assume that $XAUTHORITY is set and I hope it is true. */
+  char const *xauthpath = NULL;
+  for (char const *const *envp = &argv[argc + 1]; *envp != NULL; ++envp) {
+    static const struct com_str xauthlit = com_str_lit("XAUTHORITY=");
+    if (com_str_ext_prefixed(*envp, xauthlit)) {
+      xauthpath = &((*envp)[xauthlit.len]);
+      break;
+    }
+  }
+
+  /* Open it and parse. */
+  com_std_fd const xauthfd = com_std_open(xauthpath, COM_STD_OPEN_READ);
+  if (xauthfd < 0) {
+    com_std_print('c', "Failure to initialize X11");
+    return false;
+  }
+
+  struct com_bin xauthbin = {
+      .ctx = (void *)(int64_t)xauthfd,
+      .flags = COM_BIN_BIG_ENDIAN,
+      .interface = {.readn = fileread, .skipn = fileskip}};
+
+  /* Loop over auth entries until we find magic cookie one. */
+  struct plr_x11_auth_entry auth_entry;
+  do {
+    com_buf_clear(xauthbuf);
+    auth_entry = plr_x11_read_auth_entry(&xauthbin, xauthbuf);
+  } while (auth_entry.success && auth_entry.family != 256);
+  if (!auth_entry.success) {
+    com_std_print('c', "Failure to initialize X11");
+    return false;
+  }
+
+  com_std_close(xauthfd);
+
+  /* TODO: Check for auth_entry.name == MAGIC ? */
+  if (!plr_x11_handshake(xfd, auth_entry)) {
+    com_std_print('c', "Failure to initialize X11");
+    return false;
+  }
+
+  com_buf_free(xauthbuf);
 
   return true;
 }
 
-static uint32_t pixel_buffer[COM_CFG_WIDTH * COM_CFG_HEIGHT];
-
 // TODO: correct error handling.
-extern int plr_display_x11_main(int argc, char *argv[]) {
+extern int plr_display_x11_main(int argc, char const *const *argv) {
   char test_string[] = "what is this?";
   char *compressed_string = NULL;
   uint32_t compressed_string_sz = 0;
@@ -366,12 +475,16 @@ extern int plr_display_x11_main(int argc, char *argv[]) {
 
   com_mat_t t0 = com_mat_mul(m1, m0);
   com_mat_t t1 = com_mat_mul_reodered(com_mat_reoder(m1), m0);
-  com_fix_print(t0.a[4]);
+  com_std_print('f', t0.a[4]);
-  com_fix_print(t1.a[4]);
+  com_std_print('f', t1.a[4]);
 
   // com_vec_t v0 = com_mat_vec_project(com_mat_reoder(m0), com_vec_identity());
   // com_fix_print(v0.a[0]);
 
+#ifndef NDEBUG
+  com_test_run();
+#endif
+
   // com_fix_run_tests();
   // com_fix_run_bench();
   // com_mat_run_bench();
@@ -380,87 +493,47 @@ extern int plr_display_x11_main(int argc, char *argv[]) {
   /* Open X11 socket for communication. */
   com_std_fd const xfd = com_std_socket(false);
   if (!xfd) {
-    com_std_printf("Failure to initialize X11\n");
+    com_std_print('c', "Failure to initialize X11");
     return -1;
   }
 
   if (!com_std_connect(xfd, false, "/tmp/.X11-unix/X0")) {
-    com_std_printf("Failure to initialize X11\n");
+    com_std_print('c', "Failure to initialize X11");
     return -1;
   }
 
-  /* First we need to authenticate X11 connection, which requires some funk. */
+  if (!plr_x11_establish_connection(xfd, argc, argv)) {
-  /* We assume that $XAUTHORITY is set and I hope it is true. */
+    com_std_print('c', "Failure to initialize X11");
-  char const *xauthpath = NULL;
-  for (char **envp = &argv[argc + 1]; *envp != NULL; ++envp) {
-    static const struct com_str xauthlit = com_str_lit("XAUTHORITY=");
-    if (com_str_ext_prefixed(*envp, xauthlit)) {
-      xauthpath = &((*envp)[xauthlit.len]);
-      break;
-    }
-  }
-
-  /* Open it and parse. */
-  com_std_fd const xauthfd = com_std_open(xauthpath, COM_STD_OPEN_READ);
-  if (xauthfd < 0) {
-    com_std_printf("Failure to initialize X11\n");
     return -1;
   }
 
-  com_buf_make_on_stack(char, xauthbuf, 1024, 0);
-  struct com_bin xauthbin = {
-      .ctx = (void *)(int64_t)xauthfd,
-      .flags = COM_BIN_BIG_ENDIAN,
-      .interface = {.readn = fileread, .skipn = fileskip}};
-
-  /* Loop over auth entries until we find magic cookie one. */
-  struct plr_x11_auth_entry auth_entry;
-  do {
-    com_buf_clear(xauthbuf);
-    auth_entry = plr_x11_read_auth_entry(&xauthbin, xauthbuf);
-  } while (auth_entry.success && auth_entry.family != 256);
-  if (!auth_entry.success) {
-    com_std_printf("Failure to initialize X11\n");
-    return -1;
-  }
-
-  com_std_close(xauthfd);
-
-  /* TODO: Check for auth_entry.name == MAGIC ? */
-  if (!plr_x11_handshake(xfd, auth_entry)) {
-    com_std_printf("Failure to initialize X11\n");
-    return -1;
-  }
-
-  com_buf_free(xauthbuf);
-
   uint32_t const gc = plr_x11_create_graphical_context(xfd);
   if (!gc) {
-    com_std_printf("Failure to initialize X11\n");
+    com_std_print('c', "Failure to initialize X11");
     return -1;
   }
 
   uint32_t const win = plr_x11_create_window(xfd);
   if (!win) {
-    com_std_printf("Failure to initialize X11\n");
+    com_std_print('c', "Failure to initialize X11");
     return -1;
   }
 
   uint32_t const fb = plr_x11_create_pixmap(xfd, win);
   if (!fb) {
-    com_std_printf("Failure to initialize X11\n");
+    com_std_print('c', "Failure to initialize X11");
     return -1;
   }
 
   // Display *display = XOpenDisplay(NULL);
   // if (NULL == display) {
-  //   com_std_printf("Failed to initialize display");
+  //   com_std_print('c', "Failed to initialize display");
   //   return -1;
   // }
 
   // Window root = DefaultRootWindow(display);
   // if (None == root) {
-  //   com_std_printf("No root window found");
+  //   com_std_print('c', "No root window found");
   //   XCloseDisplay(display);
   //   return -1;
   // }
@@ -473,7 +546,7 @@ extern int plr_display_x11_main(int argc, char *argv[]) {
   //     XCreateSimpleWindow(display, root, 0, 0, WIDTH, HEIGHT, 0, 0,
   //     0xffffffff);
   // if (None == window) {
-  //   com_std_printf("Failed to create window");
+  //   com_std_print('c', "Failed to create window");
   //   XCloseDisplay(display);
   //   return -1;
   // }
@@ -507,7 +580,7 @@ extern int plr_display_x11_main(int argc, char *argv[]) {
   //                               pixel_buffer, WIDTH, HEIGHT, 32, 0);
 
   // if (!ximage) {
-  //   com_std_printf("Failed to create XImage\n");
+  //   com_std_print('c', "Failed to create XImage");
   //   // com_std_alloc(pixel_buffer, 0);
   //   XCloseDisplay(display);
   //   return -1;
@@ -547,16 +620,22 @@ extern int plr_display_x11_main(int argc, char *argv[]) {
 
   // XCloseDisplay(display);
 
-  while (1) {
+  uint8_t *fbm = plr_draw_get_pixel_buffer();
-    plr_draw_screen((uint8_t *)pixel_buffer);
+  plr_draw_screen(fbm);
 
-    if (!plr_x11_put_framebuffer(xfd, gc, win, fb, pixel_buffer)) {
+  // if (!plr_x11_put_framebuffer(xfd, gc, fb, fbm)) {
-      com_std_printf("Failure to drive X11\n");
+  //   com_std_print('c', "Failure to drive X11");
+  //   return -1;
+  // }
+
+  if (!plr_x11_show_window(xfd, win)) {
+    com_std_print('c', "Failure to drive X11");
     return -1;
   }
 
-    if (!plr_x11_show_window(xfd, win)) {
+  while (1) {
-      com_std_printf("Failure to drive X11\n");
+    if (!plr_x11_poll_errors(xfd)) {
+      com_std_print('c', "X11 closed connection");
       return -1;
     }
   }

Player/Main/cosmo.c

@@ -0,0 +1,7 @@
+#include "../../Common/std.h"
+#include "../Display/display.h"
+#include <stdio.h>
+extern int main(int argc, char const *const *argv) {
+  com_std_write(COM_STD_OUTFD, "TEST?", sizeof "TEST?" - 1);
+  return plr_display_x11_main(argc, argv);
+}

Player/Main/libc.c

@@ -1,3 +0,0 @@
-#include "../Display/display.h"
-
-extern int main(int argc, char **argv) { plr_display_x11_main(argc, argv); }

Player/Makefile

@@ -4,7 +4,15 @@ DEBUG 	?= 1 # Emit debug info.
 SQUEEZE ?= 0 # Nuclear option to save space.
 BITS 	?= 64
 RUN 	?= 0
-CFLAGS = -Wall -std=gnu99 -flto=thin -ffast-math -fno-ident -fvisibility=hidden -fuse-ld=lld
+
+ISCLANG = 1
+ifeq ($(MAKECMDGOALS),cosmo)
+	ISCLANG = 0
+endif
+CFLAGS = -Wall -std=gnu99 -ffast-math -fno-ident -fvisibility=hidden
+ifeq ($(ISCLANG), 1)
+	CFLAGS = -flto=thin -fuse-ld=lld
+endif
 ifeq ($(MAKECMDGOALS),linux)
 	CFLAGS += -fno-pic -fno-pie
 	ifeq ($(RELEASE), 1)
@@ -12,7 +20,10 @@ ifeq ($(MAKECMDGOALS),linux)
 	    		  -Wl,--optimize-bb-jumps
 	endif
 	ifeq ($(SQUEEZE), 1)
-	    CFLAGS += -Wl,-T,tiny.ld -Wl,--icf=all -Wl,--bp-compression-sort=both \
+		ifeq ($(DEBUG), 0)
+			CFLAGS += -Wl,-T,tiny.ld 
+		endif
+	    CFLAGS += -Wl,--icf=all -Wl,--bp-compression-sort=both \
 	    		  -Wl,--nmagic
 	endif
 	ifeq ($(BITS), 64)
@@ -37,32 +48,27 @@ ifeq ($(RELEASE), 1)
     		  -Wl,-mllvm,--vectorize-slp=true \
     		  -Wl,-mllvm,--enable-gvn-hoist=true
 endif
-# TODO: Just use amalgam build for all?
-DEPS = ../Common/lzw.h ../Common/mat.h ../Common/vec.h ../Common/fix.h ../Common/def.h ../Common/bits.h \
-	   ../Common/gem.h \
-	   ../Common/Timer/time.h \
-	    ./Display/display.h \
-	    ./draw.h
-SRC = ../Common/lzw.c ../Common/fix.c ../Common/mat.c ./draw.c
-LINUX = ./amalgam.c
-WASM = ./amalgam.c
-DOS = maindos.c Display/dos.c
 
-%.o: %.c $(DEPS)
-	$(CC) -c -o $@ $< $(CFLAGS)
-
-%.o: %.s $(DEPS)
-	$(CC) -c -o $@ $< $(CFLAGS)
-
-# TODO: Make it work with no stdlib.
 linux: CFLAGS += -static -mbmi2 -nostdlib
-linux: $(LINUX) | $(SRC)
+linux: ./amalgam.c
 	$(CC) -o player $^ $(CFLAGS)
 
 # Note: lld package might need to be downloaded.
 wasm: CFLAGS += --target=wasm32 -nostdlib -Wl,--no-entry -msimd128 -mbulk-memory
 wasm:
-	$(CC) -o player.wasm $(WASM) $(CFLAGS)
+	$(CC) -o player.wasm ./amalgam.c $(CFLAGS)
+
+# todo: Add -mtiny for release without debug
+cosmo: CC = ./Cosmo/bin/cosmocc -I /usr/include/X11 -mdbg
+cosmo: ./amalgam.c
+	$(CC) -o ./Out/Cosmo/player $^
+ifeq ($(RUN), 1)
+ifeq ($(DEBUG), 1)
+	gdb ./Out/Cosmo/player.com.dbg
+else 
+	./Out/Cosmo/player
+endif
+endif
 
 # https://tippach.hier-im-netz.de/wdosx
 dos: CFLAGS += -ffreestanding -m32 -fno-PIC -fno-PIE -nostdlib -nodefaultlibs
@@ -79,5 +85,6 @@ else
 	dosbox -c "mount c $(PWD)" -c "c:" -c "DOS.EXE"
 endif
 endif
+
 clean:
 	rm -r *.o ../*.o

Player/Out/Cosmo/run.cmd

@@ -0,0 +1,8 @@
+:;# amalgam of batch and shell scripts as a entry single point
+
+:; SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+:; $SCRIPT_DIR/ape $SCRIPT_DIR/player
+:; exit
+
+@echo off
+player

Player/amalgam.c

@@ -21,8 +21,12 @@
 #endif
 #include "../Common/Timer/stub.c"
 #include "./Display/x11.c"
-// #include "./Main/libc.c"
 #elif COM_DEF_TARGET == COM_DEF_TARGET_WASM
 #include "../Common/Std/stub.c"
 #include "../Common/Timer/stub.c"
+#elif COM_DEF_TARGET == COM_DEF_TARGET_COSMO
+#include "../Common/Std/libc.c"
+#include "../Common/Timer/posix.c"
+#include "./Display/x11.c"
+#include "./Main/cosmo.c"
 #endif

Player/draw.c

@@ -2,30 +2,30 @@
 #include "../Common/cfg.h"
 #include "../Common/def.h"
 #include "../Common/gem.h"
-#include "../Common/std.h"
 #include "../Common/time.h"
 #include <stdint.h>
 
-#if COM_DEF_TARGET == COM_DEF_TARGET_WASM
+static _Alignas(16) uint8_t plr_draw_buffer[COM_CFG_FBSIZE];
-uint8_t plr_draw_buffer[COM_CFG_WIDTH * COM_CFG_HEIGHT * 4];
 
 com_def_export_sym(
     "plr_draw_get_pixel_buffer") uint8_t *plr_draw_get_pixel_buffer(void) {
   return plr_draw_buffer;
 }
-#endif
 
 com_def_export_sym("plr_draw_screen") void plr_draw_screen(
     uint8_t *pixel_buffer) {
+  uint8_t *pixel_buffer_aligned = __builtin_assume_aligned(pixel_buffer, 16);
   for (int y = 0; y < COM_CFG_HEIGHT; y++) {
     for (int x = 0; x < COM_CFG_WIDTH; x++) {
-      int pixel_index = (y * COM_CFG_WIDTH + x) * 4;
+      int pixel_index = (y * COM_CFG_WIDTH + x) * COM_CFG_COLORS;
 
       // Assuming standard Little-Endian 32-bit BGRA format
-      pixel_buffer[pixel_index + 0] = (x) % 256;     // Blue
+      pixel_buffer_aligned[pixel_index + 0] = (x) % 256;     // Blue
-      pixel_buffer[pixel_index + 1] = (y) % 256;     // Green
+      pixel_buffer_aligned[pixel_index + 1] = (y) % 256;     // Green
-      pixel_buffer[pixel_index + 2] = (x + y) % 256; // Red
+      pixel_buffer_aligned[pixel_index + 2] = (x + y) % 256; // Red
-      pixel_buffer[pixel_index + 3] = 255;           // Alpha/Padding
+#if COM_CFG_COLORS == 4
+      pixel_buffer_aligned[pixel_index + 3] = 255; // Alpha/Padding
+#endif
     }
   }
 
@@ -65,7 +65,7 @@ com_def_export_sym("plr_draw_screen") void plr_draw_screen(
   // com_std_printf("%i, %i\n", p2.a[0], p2.a[1]);
   // com_std_printf("%i, %i\n", p3.a[0], p3.a[1]);
 
-  uint8_t tex[32 * 32 * 3] = {0};
+  uint8_t _Alignas(16) tex[32 * 32 * 3] = {0};
   for (int y = 0; y < 32; ++y) {
     for (int x = 0; x < 32; ++x) {
       // uint32_t idx = com_zord_order(x, y);

Player/draw.h

@@ -4,5 +4,6 @@
 #include <stdint.h>
 
 extern void plr_draw_screen(uint8_t *pixel_buffer);
+extern uint8_t *plr_draw_get_pixel_buffer(void);
 
 #endif

Player/tiny.ld

@@ -14,19 +14,19 @@ SECTIONS
     *(.text*)
   }
 
-  .rodata : {
+  .rodata ALIGN(16) : {
     *(.rodata*)
   }
 
-  .data : {
+  .data ALIGN(16) : {
     *(.data*)
   }
 
-  .shstrtab : {
+  .shstrtab ALIGN(16) : {
     *(.shstrtab*)
   }
 
-  .bss : {
+  .bss ALIGN(16) : {
     *(.bss*)
     *(COMMON)
   }