TAB doesn't work for controls anymore

[Dmitry Timoshkov]

"Gerhard W. Gruber" <sparhawk at gmx.at> wrote:

> >Simply: by writing a test program. If you can't reproduce the targeted behaviour
> >by a test, then it's very unlikely that there is an application that depends on
> >that. Otherwise your efforts are useless.
> 
> But this works only for functions that are undeocumented. How can you write a
> test program for a function that is nowhere documented? You wouldn even know
> about it untill a programm breaks because it requires it and then you have to
> do some reverese engineering for that. At least in the application to
> determine what it expects from the function.

First of all we were talking about not documented functionality, not functions.
This clearly can be done with a test application. One of the regression testing
framework primary goal is to test for that kind of things as well.

With undocumented functions I'd primarily rely on the documentation found on
the net, next - inspecting Wine relay traces with a native dll, and only as
a latest resort - look at the disassembly (in the countries where this isn't
prohibitted).

-- 
Dmitry.