ToDo List
----------------


- Need to write a "repair_cluster.pl" script that 
  modifies configuration to reflect new configuration,
  dropping and adding paths and listeners.

  This would compare the configuration computed (as in
  slonik_init_cluster) with the configuration actually found on some
  node, add "additional" bits, and drop obsolete bits.

  [Note: If slonik_drop_node and slonik_store_node are used, this may
  not be as immediately necessary, but would still be nice to have.]

- At present, the configuration generated by this set of tools is
  fairly fragile.  If just about any sort of error is made, it is
  commonly needful to drop all of the Slony schemas, thereby cleaning
  _everything_ out, and restarting the configuration process from
  scratch.  That certainly isn't ideal, and is partially preventable
  by proofreading the slonik scripts before running them, but more
  robust error-handling (and prevention) would be welcome.

- The scripts are currently not compatible with "use strict".  They
  would probably become easier to follow if they were (particularly
  with regard to figuring out where variables are defined).

- We probably don't need one script per slonik command.  Which of
  these can be combined, preferably without having to implement pages
  of arcane command-line options.

- slon_start, slon_kill, and possibly slon_watchdog should be made
  into a slon_ctl shell script, with an /etc/init.d-compatible
  wrapper.