Scennario Brainstorming, April 27th

Discussion

  • UI enhancements
    1. Tree-based view of scenarios
    2. Modal windows are a pain. Need to close all to see result.
  • Highlight
    1. Improve highlighting when path has been done multiple times.
    2. Colors for coverage
    3. Possibly thickness of path
  • Debugging?
    1. Eclipse breakpoints or tokens on path.
    2. Values of variables at certain points.
    3. Step-by-step execution
  • Linear scenario
    1. Multiple windows for different paths when forks would be interesting (nice to have).
  • Data Types
    • Mandatory:
      1. Boolean
    • Highly desired
      1. Integers (with arithmetic operations)
      2. Enumerated types
    • Would be nice
      1. Arrays
    • Would be nice but not important
      1. Real values (depends on what you want to compute)
      2. Objects? Structures? Not goal of UCMs but might be iteresting if you want to test something else.
  • Responsibilities
    • Modify state variables
    • Sequential operations?
    • If structure?
    • Can invoke environment programs?
  • Need to support different styles
    1. From very structure style using stubs to many paths that are connected.
  • Visual indicatiosn: Start points have additional visual clues that its precondition is set to be "the map must reach a certain point".
  • Traversal should be simple
    • Have syntactic sugar to simplify the rest.
  • Long discussion on what/how should be expressed in UCMs with regards to external events such as the user answers the phone.
    • Yong: Should not be in UCMs.
    • Gunter: When the interaction is very important, it must be included. UCMs are the starting point; you want the whole view where you can do tests on it.
    • Conditional visualization / pruning of these cases; auto-generation of branches for events.
  • Inverse scenarios
    1. Find scenarios from paths (Jamel: Slicing)
    2. Need to traverse both ways
  • Non-determinism? Exclusive forks? These questions cannot be answered. The traversal mechanism must be extensible to support various styles of traversals of the system.
  • Coverage visualization. Right now, it is very difficult to follow the path where there is a deep map structure. Furthermore, it is even harder to find what has been covered; the main goal is sometimes not to find what has been covered but rather to find what has not been covered.
  • We don't want individual scenarios anymore. We wish to think of scenarios as building blocks that can be re-used / modularized so that one change doesn't have to be replicated in tons of scenarios.
  • Because of testing wishes, must define post-conditions and end points that must be reached. (Initial feeling is that we should specify the exact set of end points that are reached, but we may want the flexibility to specify only a subset - or the ability to specify an expression which must be satisfied, e.g. EndPointA or EndPointB).
Topic revision: r2 - 04 May 2006, GunterMussbacher
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback