Honestly, 2.5 years is way too long to wait for a NIC driver. The actual amount of code that has to be written isn't that great, and open source Linux drivers could probably be used for reference.
In all fairness, having a Linux model doesn't always offer as much utility as people think.
Of all the NG OS', AROS seems to be the operating system that benefits most from Linux examples.
The PPC OS' usually require tighter more streamlined solutions.
But this difference is why I favor NG OS', as Linux uses an ever growing brute force model, while NG OS' descend from a 68K OS where resources were much tighter and coding efficiency had to be maximized.
And my own coding is suited to this as I always wrote for modularity, breaking down tasks into separate sub parts.
What worked well once to limit resource demands on 68K OS' now serves well in SMP OS' to maximize threading.
Also, I think we've had some really creative coders in the Amiga community, and some of them are still here turning out improved solutions.
Further, I just exchanged a message with Ralph Schmidt today, and I made a point of thanking him for his use of a micro kernel in MorphOS. With no apologies to Torvalds, I still think this is a more efficient and stable way to base an OS (unlike the massive monolithic kernel Linux is based on - where the kernel alone is larger than AmigaOS in its entirety).
We have a history of thinking outside of the box, basically descended from our heritage with the first commonly available multimedia computer.
And even some of our dead end were pretty cool (like I would have loved to see the QNX based fork of AmigaOS continue to be developed).
SO...what's the point of this post? Well, Linux models reflect their inefficient UNIX heritage. And as someone who moved from 68K Xenix based systems to more efficient 68K OS' that were being promoted for process control applications, I KNOW there are better ways to do things than the mainstream standards.
Basically, just because Windows, OSX/MacOS and Linux (at a very distant third) dominate the market, it DOESN'T mean we have to model our approaches on these OS'.
Sometimes different can be better.