restructed all the files + uci done now

also broke the symbol analyser in the process, i didn't help much, i had
to pull out the pen & paper, but it made a cool graph never the less

multithreading was a pain! but it got it in the end

fixed the structs naming convention, if it's used outside by other
files, it's typedeffed, otherwise it's not

fixed the recursive imports issued i had in go_args*, where i was
casting it to void* in uci to make it work

now response.c doesn't deal with ipc, really happy with that change
now every single thing has its own responsibility

1 files changed, 91 insertions, 0 deletions

diff --git a/src/engine/threads.c b/src/engine/threads.c
new file mode 100644
index 0000000..5afe155
--- /dev/null
+++ b/src/engine/threads.c
@@ -0,0 +1,91 @@
+#define ENGINE_THREADS_INTERNAL
+#include "engine/threads.h"
+#include <string.h>
+#include <stdlib.h>
+#include <assert.h>
+#include <stdio.h>
+
+void threads_init(struct threads *handler, size_t capacity) {
+  // maybe this will be a speed up?
+  // why am i micro optimizing here??????
+  // idk i just watched the eskil steenberg video about UB
+  memset(handler, 0, sizeof(*handler));
+
+  handler->threads = malloc(sizeof(*handler->threads) * capacity);
+  handler->args = malloc(sizeof(*handler->args) * capacity);
+  handler->capacity = capacity;
+  handler->count = 0;
+  handler->running_count = 0;
+  handler->engine_messages = NULL;
+  handler->go_args = NULL;
+  handler->position = NULL;
+}
+
+void threads_deinit(struct threads *handler) {
+  assert(handler->running_count == 0);
+  free(handler->threads);
+  free(handler->args);
+}
+
+void threads_reserve(struct threads *handler, size_t capacity) {
+  if (handler->capacity >= capacity) return;
+
+  handler->threads = realloc(
+    handler->threads,
+    sizeof(*handler->threads) * capacity
+  );
+  handler->args = realloc(
+    handler->args,
+    sizeof(*handler->args) * capacity
+  );
+  handler->capacity = capacity;
+}
+
+void threads_go(struct threads *handler) {
+  int running_count = handler->running_count;
+  for (int i = 0; i < running_count; i++) {
+    __threads_args_init(handler->args + i, handler, i);
+    pthread_create(
+      handler->threads + i,
+      NULL,
+      engine_thread,
+      handler->args + i
+    );
+  }
+}
+
+bool threads_go_loop(struct threads *handler) {
+  // idk maybe do something while waiting for calculations to flow in
+}
+
+bool threads_all_done(struct threads *handler) {
+  int running_count = handler->running_count;
+  assert(running_count > 0);
+  for (int i = 0; i < running_count; i++) {
+    if (atomic_load(&handler->args[i].finished) == 1) continue;
+    return false;
+  }
+  return true;
+}
+
+void threads_cleanup(struct threads *handler) {
+  int running_count = handler->running_count;
+  for (int i = 0; i < running_count; i++) {
+    __threads_args_init(handler->args + i, handler, i);
+    pthread_cancel(handler->threads[i]);
+  }
+  handler->running_count = 0;
+}
+
+void __threads_args_init(
+  struct thread_args* args,
+  struct threads *handler,
+  int i
+) {
+  atomic_init(&args->stop, 0);
+  atomic_init(&args->finished, 0);
+  args->id = i;
+  args->position = handler->position;
+  args->go_args = handler->go_args;
+  args->message = handler->engine_messages->data[i];
+}