Error message
In the message, %s marks a value that the engine fills in when it raises the error.
The story
You handed a network builtin something it cannot use, and the engine stopped before touching the network at all. Network rows work with handles. When you open a server, you get back a map, and inside it a listening socket. Next request and close both need that map. When a request comes in, you get another map, with an open connection inside, and respond needs that one. If you pass something else, say a number, a text, or the wrong map, the row has nothing real to work with. The same goes for an HTTP status code outside one hundred to five ninety nine: it is not a valid status. Why is the engine so careful? Because writing to a descriptor it did not open could damage something unrelated. So it refuses first, and names the row and the mistake. The fix: keep the map that server open gave you, pass it to the next call exactly as received, and use only real status codes. Right handle, right row, and the conversation with the network flows.
Text from the Rush engine's error catalogue.