I am automating an e-commerce site www.music123.com and I was looking forward to hearing your suggestions. In my experience, I generally have 1 test (with 1 action and 1 shared repository) for each area of functionality. Functionality can be categorized into search, my account, checkout, payment etc. At the end of it, I might have about 6 tests for the entire project which I would run in a batch job using windows scheduler.
I was reading the QTP product documentation and it seems to suggest 1 Test with several actions (each action with it's own shared repository) in it's Mercury Demo Tours project.
So my question is, if there's a general rule of thumb for easier maintainability of scripts. Should I have 1 Test with several actions for each functionality or several tests each with a single action and 1 shared repository?
If the post isn't sufficiently clear, I'll be happy to clarify.
It entirely depends on the approach to automatation. Its easy to maintain and debug, if you have all the functions written in different actions. Then use a Shared Object Repository, instead of each action having its own Object Repository. Then Including these external actions in the code would again make them reuseable.
Its all about how you get to debug and maintain the existing scripts.