From the moment Intel Macs became available, running Windows apps on Macs has been a topic of interest in the Mac community. Suddenly, Macs became the ultimate machines for those of us interested in running multiple operating systems. Sure, Linux was always available for PowerPC Macs--at least several distros--but running Windows at satisfactory speeds was always a challenge. With the advent of Intel Macs, there are choices aplenty. We can (1) dual-boot with Bootcamp; we can (2) run Windows with virtualization software like Parallels or VMWare, which run at near-native speeds, significantly faster than the emulation we did on our PowerPC Macs; and now we can (3) run Windows apps without Windows using CodeWeavers' CrossOver.
Those solutions are geared to respectively increase convenience. Bootcamp requires an inconvenient reboot. Then virtualization software allows us to boot up Windows without rebooting our Mac, and we can even use Mac apps and Windows apps simultaneously albeit with Windows in an encapsulated environment. Finally, CrossOver aims to take the convenience to the next level by eliminating Windows and enabling Mac OS X to execute Windows apps!
One article at LinuxWorld called CrossOver Office "the best way to do the wrong thing". I contend that CrossOver is the worst way to do the wrong thing. If I have to run a Windows app on my Mac, I certainly want it to run as smoothly as possible, just like the rest of my Mac experience, and I want it to run like the developer intended. Largely, dual-booting and virtualization don't compromise the behavior of the app I'm running. When you run your app with CrossOver, however, you don't know what kind of performance you'll get. Your app may just die; it may run but be full of bugs. The reports coming from the web show that your experience will be very hit-or-miss.
Don't get me wrong; I'm not knocking the WINE project, which is the basis for the CrossOver codebase. A project to port the Windows APIs to another platform, although ambitious, is fascinating and, frankly, perfect as an opensource project. The problem I have is with the attempt to commercialize this technology that is great as a free resource but destined to never deliver a level of quality that befits a commercial product.
Why? There are so many variables and pitfalls to porting an API that the technology will never be able to work for even a large percentage of Windows apps, let alone all or the majority. And the technology can continue to be refined, only to see a Windows upgrade completely shatter the compatibility of future apps, and the development process of tweaking the port starts all over again. And from an opensource perspective, that's fine. That's the strength of an opensource initiative. It's not such a great model for a commercial product.
What would you rather do? Buy CrossOver for $59 and be able to run only a few apps with it, and perhaps with a few bugs at that, or buy Parallels for $79 and be able to run practically all Windows apps with it, nearly bug-free? Of course, the price differential increases if you need to purchase a copy of Windows. Nevertheless, your experience will be infinitely more reliable if you use a virtualization solution like Parallels.
For the tech geek who likes tinkering with new software, this solution is worth a gander, especially while the free public beta is available. However, if you just need to get down to business, I recommend sticking with virtualization.