> First off, in reality there is no such thing as a stable kernel ABI in any mainstream > operating system out there.
Solaris and AIX manage to have very stable kernel ABIs.
> So in your example, your hardware wouldn't have survived either the transition from > XP to XP64, or XP to Vista. And not to Windows 7 either which by default requires > drivers to be signed.
Did you read the post you replied to? No claims were made about Windows. Strawman argument is strawman.
> Had you instead open sourced your driver it could have been ported to all the above, > and had you submitted it for inclusion to the Linux kernel it would have been > maintained by the kernel devs and kept functioning across any kernel abi/api changes.
Once again, read the fucking post you replied to: it's assumed the driver is open-source but that doesn't magically create maintainers for it; the Linux kernel tree is not the place for drivers for every niche product; the kernel devs may be able to keep a driver building across kernel releases but they have no way of testing it to ensure it's functional.
So in summary, you pull a strawman and then produce the arguments that the post your responded to already addressed. As long as Linux fans ignore the problems, they won't be solved.