I have just been tasked (sentenced) to define and design a new automated testing system that basically any functional tester can use without a whole bunch of technical expertise other than with the application to be tested (a huge governmental money handling system using 3270 screens, no GUI).
What we already have:
1. QA Center (with only QARun really used);
2. A stand alone GUI application using Oracle database to maintain requirements, etc.
3. Neither is connected. Due to political considerations, the stand alone will not go away.
My thoughts are:
1. Convert everything to QA center - I think this is not desirable because of the manner of storage of requirements, i.e., Word Document.
2. Interface the stand alone to QA run with new code.
3. Interface the stand alone to QA Director.
4. Create a stand alone to do everything.
5. Quit and retire to my farm and raise honeybees.
All of this is also in furterance of the effort to go from thousands of individual "learned" scripts to a data driven model.
Any and all help is greatly appreciated. Also bear in mind that #5 is starting to look pretty good. ;D
I'll give you a little tip (if you're looking to go to data-driven testing).
The most important thing is to know exactly *WHERE YOU ARE* in your process. If you don't know where you are, you don't know what you can do.
Remembering what you typed is easy. Verifying that what you typed on page 1 shows up on page 10 is also easy.
1) Where are you?
2) Where can you go from there/what can you do?
If you can nail those two problems, then you will get much closer to your final goal. Remember, you can't automate everything (automate the 90% cases first), and you will need "hooks" or ways to "break out" of whatever framework you end up working with.
One other thing- you are focused on improving/maintaining the quality of your application. It shouldn't be your goal to have everything perfect, and be able to automate and verify every last detail. Verify the stuff that breaks most often (look at historical bug data), and what has the most impact if it were to go wrong.