SOC Automated Windows Testing Project

[Andrew Tridgell]

Jim,

 > The main problem with this approach is if you try to do something that can 
 > only be done in a GUI, or if you're trying to test a GUI app at all.  I'm 
 > not saying it shouldn't be done, but just that it's a boundary you can't 
 > cross without doing a graphical login.

I think both proposed approaches have problems with testing GUI
apps. I suggested to Brad that he concentrate initially on getting
non-GUI testing working, then only look at methods for GUI testing if
he has time.

If he does get time, then it looks like there are quite a few windows
GUI scripting systems freely available (just from a quick google
search). I'm not sure how we'd trigger these from the Unix side of
'make test'. I guess we could connect with rdesktop and setup a logon
script that starts a series of GUI scripts. It might be tricky to work
out whether the app has behaved correctly though.

Cheers, Tridge