On Tue, 23 Aug 2005 02:00:32 GMT TokaMundo wrote: | On 19 Aug 2005 22:16:34 GMT, snipped-for-privacy@ipal.net Gave us: | |>On Thu, 18 Aug 2005 21:36:05 GMT TokaMundo wrote: |>
|>| Using a client that allows one to press "send" over and over again |>| calls for changing the client to one that actually had more than two |>| days programming time coded into it. Your interpretive powers lack a |>| bit as well, considering your new take on those that responded to you. |>
|>I sure as hell would NEVER use a client that refused to let me retry a |>post when the first attempt failed. | | It didn't fail, dipshit. If one is ON a sever, one is already | LOGGED in. One would have no problem sending one's message with a | single click of the send button. Seems your knowledge of human | interface lacks a bit. It's a poorly programmed client that allows | something so stupid. I can RESEND my posts, but I only get to send | ONCE. You are clueless.
You're doing way too much of that weed. It has destroyed your brain (assuming you ever had one to begin with).
If the TCP connection is LOST during the sending, before the client gets back a positive acknowledgement, the correct course of action for a client is to retain the message. Whether it QUEUES the message for AUTOMATIC reposting when a new connection is established, or requires the user to click POST again is part of the client design choice. But that has nothing to do with the original cause of failure.
The client design that would repost automatically is also a dangerous one. It is very possible (and indeed really happened in this case) that a bad condition in a network or server really can result in re-submission. If the TCP connection to the NNTP port is lost, the only way to get the message posted is to establish a new connection (while in the mean time, the server thought the prior posting had completed). In this NEW connection, everything starts all over again since the original connection state is not a part of the new connection. It even requires logging in all over again where login is required (most places now days).
I take it you have never done protocol design and implementation. I have, and I know the complex issues involved. A lot has been learned since the first protocols for TCP/IP were designed.
Servers really can get bogged down. Servers really can have a flaky network interface card that causes intermitted or rapid network failures. If you believe these things can't ever happen, you obviously haven't worked in any high tech environment.
So get a clue, pickle brain.