Do not expect it to make ReactOS magically better. Some things will be better, others will not. Whether you think it is better depends on which applications you use.
Think of this as using Wine's versions of the presentation layer (user32/gdi32). This creates the user controls and allows interacting with a bunch of stuff.
User32 manages appearance and input, and it typically tells Gdi32 what to draw and where. If User32 knows what to do, but GDI32 does not know how to draw it (or calculates incorrectly) then User32 looks like it does not know what it is doing. Of course USer32 may have incorrect or missing implementations also. The functions are named rather clearly, so even if you don't know a lot about Windows, this list might help see what they do. User32 has a lot of functions involving the words keyboard, window, message, menus, and piles of stuff about resources (icons, cursors, colors). When you click or copy or paste or type, almost 100% of the time user32 is involved before anything else is.
http://www.webtropy.com/articles/art9-2 ... user32.dll
http://www.webtropy.com/articles/art9-2 ... =gdi32.dll
or maybe this page, I'm not sure but the intent is to describe how a virus might use the function.
http://www.pc1news.com/virus/module-user32-dll-54.html
The only practical reason for integrating Arwinss is to figure out where bugs actually are. Wine is not perfect, but it works with a lot of popular applications. So ReactOS will appear to work better in some ways. Here is the problem. Wine expects to run on X11, and Arwinss pretends to be X11. It intercepts X11 calls and re-directs them to the ReactOS kernel. Of course, Arwinss is not actually X11, and it may have its own bugs. So it will probably not be exactly as good as Wine.
It is not within the goal of ReactOS to host Wine in this way - ReactOS is supposed to be as close to Windows as possible. It gives the appearance of compatibility only. Some people may say what is the difference, as long as it is under the hood and applications work as expected? You make up your own answer, it has been discussed and I guarantee your opinion has been stated before by someone else, and there is probably at least once developer who disagrees vehemently.
As a developer, if you are not working on user32/gdi32, you can use Arwinss as a way to have something to build on top of. It is temporary support so that you can build from the top down, while waiting for the base to be built up. As a user, it might be nice to try a beta release with a replacement subsystem and have different things working. A lot of large projects have had parallel development going on, and either the best man wins or both make progress. Either way, it works out for the best.
It allows one more way for the code that *can* be shared between ReactOS and Wine to be tested.
It would have to be either some system-wide switch (use ARWINSS DLLs now instead of the usual ones) or an application profile manager with per-app settings.
