Last commit for todo.txt: 6c053c3cbe929ebe4d91d24b3c68f3e7da4696ee

nothing important

jg253 [2002-03-12 22:40:14]
nothing important
  1. TODO:
  2. - shutdown on Processor should do some cleanup:
  3. + inform peers that it is going down
  4. + maybe if some task is still running, refuse to shutdown, etc
  5. - what to do if receive request to run a task that is already executing, but
  6. hasn't completed ... ie. is an 'endangered task'
  7. - need to fix the versioning to have a pre-execution, mid-execution and
  8. post-execution version
  9. - fix problem when a remoteTaskProcessor is really local (make it be a local
  10. call instead of using worklets)
  11. - global logging + standardize error messages
  12. - hierarchy for net/*containers -or- unify net/*containers into one
  13. container
  14. - standardize some of the arguments and names of files
  15. - make names in VTransportContainer more logical
  16. - get rid of the ugliness in which the handles are being used. we
  17. should probably use some other class to invoke the transport
  18. layer cleanly.
  19. - swap the semantics meaning between valid and ping! It's currently
  20. the opposite of what it should be.
  21. - get rid of 2 objects in Version object
  22. - fix service discovery aspect so that we don't learn about each service more than once (as happens now). It's a small bug that doesn't affect the "working" nature of the final product, but it's still a bug.
  23.  
  24. DONE:
  25. *- get Replicators working by default in the ProcessorMain, fix STOP-TASK,
  26. use reflection to obtain processor-TYPE
  27. *- fix the replication of data (such as name, port and hostname) all
  28. over the place.
  29. *- differentiate between valid and ping (right now, they are treated
  30. the same in the net layer, which is dumb)
  31. *- stop task ... done, I think ... shutdown cloud ... done, but need
  32. cleanup
  33. *- fix XML schema to be more reflective of the XML we are actually using
  34. *- WVM should be created before the ServiceHost_Serv ++ NRLProcessor
  35. coz it needs the rmiregistry ... or in the very least, the
  36. ServiceHost_Serv should be created *after* the WVM .... might wanna
  37. make this a singleton ServiceHost_Serv, btw
  38. *- need to either change the RMI name for the ServiceHost_Serv to
  39. <taskName>-serviceHost, or not use RMI at all here
  40. *- create a different ServiceHost_Serv/Scheduler_Serv for every invocation
  41. of executeTaskLocal in NRLProcessor. Actually, there is only a single
  42. instance of these objects, but they've been modified to support multiple
  43. task processing
  44. *- what to do when receive multiple requests to kill a certain task? well,
  45. just add all requests to the _stoppedTasks Set so that can ignore the result
  46. data of this task
  47.  
  48.