I see. Well, what I really need to get working on Linux are WinModems, so I don't think performance is a problem there . But, in a driver that needs real-time response, it will be a problem. If someone wants to start this project, then we need somewhere to collaborate and things like that, SourceForge maybe? If you want to start this, then we can't just keep on talking about it, we actually have to start.

Hi,

I had an idea a while back that I had thought would work great for software I think that it can work great for drivers too.

It would be written in ASM. You know how some software can tell what your modem is doing!!! For example detecting what password and user name it's is using to log in to a account!!!

Well a driver and a software program that does that with software and drivers. It will detect what the programing of the software and the driver is. Sort of like how some hackers gets the programming out of chips but it would not reveal the programming it would just adapt the programing to run on each OS individually. No linking just running properly on what ever OS as if it was written for the OS and in a way it was written for the OS. The adaptation results of the programing would be closed sourced for legal reasons.

Picture this!!! clicking on an exe file for the windows driver or the windows software and it installs it as if it was made for the different OS by cleverly re-engineering the programing.

NDIS is a standard for network drivers that can be used in DOS and Windows. For sound cards, graphic cards, modems, scanners, printers, tablets, etc to be used across all plateforms (PC, MAC, SPARC, etc) there have to be standards. The computer industry does not like standards to well because it costs money. Standards creates a mess of things if developers of the standard are not carefull. ReactOS is already using some Windows drivers but what I read is it is unstable. Linux can not do this because it is like putting a round peg in square hole.

There are plenty of documentation about the Linux kernel. Windows documentation is actually nothing if you compare it to Linux documentation. The WINE developers have to improvise many of Windows API layers. IMHO, the WINE project is a lost cause for Linux users. I recommend using VMware or wait until Xen source to include virtual hardware extensions to use Windows and Linux natively at the same time.

Actually even with vmware, universal windows driver support is'nt quite a reality. Vmware does'nt provide direct hardware access to pci cards I've noticed. VMware doesn't even see my raid card, or my graphics card. Besides, even if it did, who wants the performance hit of running windows on top of linux? I say our best hope of getting more hardware/vendor support is EFI. It would be not only beneficial to linux but to any free OS. It would even by beneficial to users of a commericial os. Imagine being able to plug in any device without worrying a damn about configuration or drivers. Its compelling. But there is a group that won't benifit from this at all. You can be sure that their next OS release won't support this, thus undermining the effort of EFI and effectively squashing the hope of any univerisal driver api.