Add TODOs

llama_proxy_man/src/main.rs

@@ -84,6 +84,10 @@ struct SharedState {
 
 #[tokio::main]
 async fn main() {
+    // TODO add autostart of models based on config
+    // abstract starting logic out of handler for this to allow seperate calls to start
+    // maybe add to SharedState & LLamaInstance ?
+
     initialize_logger();
     // Read and parse the YAML configuration
     let config_str = std::fs::read_to_string("config.yaml").expect("Failed to read config.yaml");
@@ -264,9 +268,10 @@ async fn handle_request(
             cmd.kill_on_drop(true);
             cmd.envs(model_config.env.clone());
             cmd.args(&args);
+            // TODO use openport crate via pick_random_unused_port for determining these
             cmd.arg("--port");
             cmd.arg(format!("{}", model_config.internal_port));
-            cmd.stdout(Stdio::null()).stderr(Stdio::null());
+            cmd.stdout(Stdio::null()).stderr(Stdio::null()); // TODO save output and allow retrieval via api
 
             tracing::info!("Starting llama-server with {:?}", cmd);
             let process = Arc::new(Mutex::new(