[Chandler-dev] Improving Functional Tests (meeting notes)

[Philippe Bossut]

Hi,

We had a meeting this afternoon discussing how we could improve Chandler 
Functional Tests. These are my short notes.

Present: John, Mikeal, Bryan, Dan, Aparna, Philippe

Objectives:
- move toward a system which is more simple to maintain
- have a system that devs like to use (i.e. write more functional tests)
- suppress the productivity sink that functional tests fixing is right now

Ideas:
- Makes the way we can find out why a functional test fails easier:
   * Script is accessing things by name by trying too hard: we should 
simplify this so that failures don't propagate
   * Script writing is too implementation specific (breaks easily), 
instead use a naming convention we have
   * Validation: whenever we can (and it makes sense) check the repo 
(easier), possible to refer to the UI elements by name
   * Proposes to use a more direct way to point to items and methods so 
that we avoid lots of this and get break points closer to the failing point
   * Problem with Dialogs: we need a different plan (we have IDs but no 
names for them)
   * Problem with focus

- Rewrite QATestLib:
   * Will be more than 2 weeks for sure
   * John will help QA folks for 2 weeks

- Modify the way we run tests so that we run them separately first then 
together (instead of just together as we do today)

- Makes the writing of functional tests easier:
   * Try to implement a recording mode, will get a simple Python code, 
will just have to add verifications

Plan:
- John starting on this project Monday 27
- Will start on writing on writting a simple test (test all day) with 
his new idea
- Will present this to the QA people and we'll iterate from that
- John will get back to alpha5 by Dec 8th

Please feel free to add precision to these short notes.

Cheers,
- Philippe