From bd2ad340e61be08b3cfb615c94cbfb708617eb59 Mon Sep 17 00:00:00 2001
From: Runxi Yu <me@runxiyu.org>
Date: Mon, 03 Mar 2025 21:34:51 +0800
Subject: [PATCH] hooks: Comment on how using the first byte as return doesn't make sense

---
 git_hooks_client/git_hooks_client.c | 16 +++++++++++-----

diff --git a/git_hooks_client/git_hooks_client.c b/git_hooks_client/git_hooks_client.c
index 46fd02a4c639a2b9826a244ab6c2d7e62ed1219e..fc49d915c8f9db46a760b71659c45a7f3c7c88f4 100644
--- a/git_hooks_client/git_hooks_client.c
+++ b/git_hooks_client/git_hooks_client.c
@@ -177,11 +177,17 @@ 		close(sock);
 		return EXIT_FAILURE;
 	}
 
-	/*
-	 * The first byte of the response from the UNIX domain socket is the
-	 * status code. We read it and record it as our return value.
-	 */
-	char status_buf[1];
+        /*
+         * The first byte of the response from the UNIX domain socket is the
+         * status code. We read it and record it as our return value.
+         *
+         * FIXME: It doesn't make sense to require the return value to be
+         * sent before the log message. However, if we were to keep splicing,
+         * it's difficult to get the last byte before EOF. Perhaps we could
+         * hack together some sort of OOB message or ancillary data, or perhaps
+         * even use signals.
+         */
+        char status_buf[1];
 	ssize_t bytes_read = read(sock, status_buf, 1);
 	switch (bytes_read) {
 	case -1:

-- 
2.48.1