TODO (0.7a)

This file is a combination of a conventional TODO file and my ideas for possible future developments of the system.

• Auto-update of the HOST_COMMANDS in the host list? : Failed commands will be removed from the list and logged.
• System responses etc. (unknown queries) (See also above). This really is necessary for a secure intermud mail system.
• Remote list updating?
• Reduce the amount of call_out's used ??? :
  Timeout incoming_packets and received_ids with hb instead of call_out. Only use call_out for externally visible calls, ie. timeouts (to prevent problems with eval cost). Perhaps use hb in all cases and then call external objects using call_out.
• "Fake Mud" checking should be done against definitions in udp.h, not date from the host list.
• The ftp protocol/reliability needs to be improved.
• Implement INETD_SAVE_FILE: file that hosts is stored in to allow last access data etc. to be retained.
• Should a special mudname "*" be supported by send_udp() to make commands such as channel commands more efficient?
• Should send_udp() be overloaded to accept an array of mudnames?
• Implement special "application" hosts ??? :
  INETD_APPS file containing applications (stored separately from regular hosts so they are not queried by query:hosts), which are referred to as "app:name" via send_udp(), or maybe a seperate function providing a more appropriate protocol ?
• When a new mud is detected, a system packet should be sent to all other muds, notifying them? At the moment a new mud may never be detected at some sites which are not listed on the new muds host list.
• Add ierror.h file to allow users to configure their own error messages. Or perhaps name it inetd.h and move all the inetd configuration options to it.
• New-style packet encoding means the DATA field is effectively obsolete. Change udp data-handling functions to use 2 arguments for intermud data: (1) the "header" (a mapping as-per normal, excluding the DATA field). (2) The "data" or "body" (what would have been in the DATA field)