http://bugs.winehq.org/show_bug.cgi?id=5693
--- Comment #10 from Jeff Zaroyko jeffz@jeffz.name 2008-11-21 20:35:46 --- (In reply to comment #8)
Cool! Can you write a conformance test to automate what you tested manually? Then I bet Alexandre would accept your patch.
I have a test that shows what I said above, but the reason I think it's incorrect is, what looks like a bad parameter could be a near pointer?
Since this is a win32 app using a win16 message, the tabs parameter which looks like an int in 32bit could really just be the offset to be used with a selector?
Even though the type of the tabs parameter is LPINT, is it possible that windows accepts a PINT as well?
I don't know much about win32 <-> win16 interop, but I'm guessing some sort of thunk is required if this is true. Need to either write a testcase for this theory or debug the app more intensively.