James G. Sack (jim) wrote:
In one sense, it's not quite fair to compare wireless difficulties with wired, because the wireless case involves more complications -- radio (short for several of those complications), modes, and auth/security.
Hogwash. All of that stuff is quite capable of being buried behind a uniform API. Every wireless card needs to perform the same set of operations *from a user perspective*. The fact that those operations have to toggle hardware specifically is the responsibility of the driver.
As another observation, I think there was, in fact, a long period of hardware reverse-engineering and catch-up in ethernet adapters before Becker (somewhat) single-handedly standardized a lot of it.
And his dogged persistence to keep the Tulip driver working no matter what meant that his driver effectively became the de facto API. People who were writing a new ethernet driver knew that there was pressure to keep the API that the Tulip driver used stable, so they wrote to that API since it was less likely to break.
-a -- KPLUG-List@kernel-panic.org http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
