Skip to content

OCS3 Prototype Plan, Kinda

Jump to bottom
Rob Norris edited this page Jun 7, 2016 · 6 revisions

Ok the idea is to do this in two steps, so we can push off the UI work until we have Carlos, and to leverage the lessons he's learning right now working on Seqexec. We are confident we can get something approaching Stage 1 done by the end of the year if we work on it mostly full-time.

Stage 1

The idea here is to build out

  • a model that supports core features and doesn't preclude wish-list features
  • a server design that’s simple and maintainable and gives high confidence of correctness
  • minimal implementations of
    • programs
    • observations
    • targets and guiding
    • sequences
  • support for instruments that are interesting in different ways
    • F2
    • GMOS
    • GHOST
    • NIRI
  • remote server monitoring/logging
  • authentication and authorization
  • command-line interface for operations on the above

Stage 2

The idea here is:

  • more model features (TBD)
  • background processes (AGS, model checking, staff/user notifications)
  • staging/deployment/monitoring/failover
  • answer some client design questions
    • how does information flow between server and client?
    • how do we handle concurrency and contention?
    • how do we ensure that the client's view is consistent with reality?
    • how do we implement the TPE in Javascript?
  • a web-based UI supporting
    • basic program/observation/target/sequence editing
    • TPE with AGS

Clone this wiki locally