+ * Revision 1.245 2009/04/24 15:29:43 fabiankeil
+ * Allow to limit the number of of client connections.
+ *
+ * Revision 1.244 2009/04/17 11:34:34 fabiankeil
+ * Style cosmetics for the IPv6 code.
+ *
+ * Revision 1.243 2009/04/17 11:27:49 fabiankeil
+ * Petr Pisar's privoxy-3.0.12-ipv6-3.diff.
+ *
+ * Revision 1.242 2009/04/11 10:44:47 fabiankeil
+ * Update a comment. We're not in Kansas anymore.
+ *
+ * Revision 1.241 2009/04/11 10:37:23 fabiankeil
+ * When dropping connections due to ACL, don't leak csp->ip_addr_str.
+ *
+ * Revision 1.240 2009/04/09 10:12:54 fabiankeil
+ * Fix two cases in which an invalid server response would result
+ * in the client connection being closed without sending an error
+ * message first.
+ *
+ * Revision 1.239 2009/04/07 11:43:50 fabiankeil
+ * If the server rudely resets the connection directly after sending the
+ * headers, pass the mess to the client instead of sending an incorrect
+ * connect-failed message. Fixes #2698674 reported by mybugaccount.
+ *
+ * Revision 1.238 2009/03/27 14:42:30 fabiankeil
+ * Correct the status code for CONNECTION_TIMEOUT_RESPONSE.
+ *
+ * Revision 1.237 2009/03/27 14:32:04 fabiankeil
+ * If spawning a child in listen_loop() fails, send a real
+ * HTTP response to the client and continue listening for
+ * new connections without artificial delay.
+ *
+ * Revision 1.236 2009/03/25 17:30:24 fabiankeil
+ * In serve(), keep the client socket open until we marked the
+ * server socket as unused. This should increase the chances
+ * that we reuse the connection for the client's next request
+ * to the same destination.
+ *
+ * Revision 1.235 2009/03/18 21:01:20 fabiankeil
+ * Comment fix. Spotted by Roland.
+ *
+ * Revision 1.234 2009/03/18 20:48:42 fabiankeil
+ * If the --no-daemon option is used, enable LOG_LEVEL_INFO
+ * before the config file has been parsed (as we always did).
+ *
+ * Revision 1.233 2009/03/13 14:10:07 fabiankeil
+ * Fix some more harmless warnings on amd64.
+ *
+ * Revision 1.232 2009/03/08 19:29:16 fabiankeil
+ * Reinitialize the timeout structure every time before passing
+ * it to select(). Apparently some implementations mess with it.
+ * Probably fixes #2669131 reported by cyberpatrol.
+ *
+ * Revision 1.231 2009/03/08 14:19:23 fabiankeil
+ * Fix justified (but harmless) compiler warnings
+ * on platforms where sizeof(int) < sizeof(long).
+ *
+ * Revision 1.230 2009/03/07 13:09:17 fabiankeil
+ * Change csp->expected_content and_csp->expected_content_length from
+ * size_t to unsigned long long to reduce the likelihood of integer
+ * overflows that would let us close the connection prematurely.
+ * Bug found while investigating #2669131, reported by cyberpatrol.
+ *