In message <5ge1vhF3g1jjnU2 / mid.individual.net>, Robert Klemme writes:
>Why?  Please explain!

BSD tends to be better engineered.  Look at it this way:

Imagine that a driver were to print an error message.  What would be the
chances that the message would unambiguously identify the driver?

Imagine that a device has been probed.  Will the probe message tell you
what the device is and give you a good hint as to the name of the driver?
Will the message give you information about the device?

I've just spent a week buried in Linux kernel code.  The core Linux code is
pretty good these days, but it would benefit a lot from more standardized and
consistent kernel messages.  To be fair, some of the code I'm working with
is vendor code that isn't in (or likely to be added to) the mainline kernel,
and that code is usually the worst.

They're philosophically different.  In general, I prefer the BSD approach,
which places greater weight on developers conforming their code a bit to
the kernel's style and standards at all levels, including displayed messages.

-s