>   If you are run the uname system call from a program that is emulating
>   linux, for example, it should return "linux" -- not netbsd -- as the os.
>
>
>why?

Uh, for the reasons stated earlier in the thread...

some commercial apps come with their own ``MI'' script wrappers which
customize the environment, &c based on the host OS, which is obtained
via uname(8).  Returning "NetBSD" for uname(2) tends to break these
commercial apps.  Supporting commercial apps is one good reason for
using an emulation.

Yeah, someone who wants to can hack such scripts so they work on
NetBSD, but that's not Bill's point. Which is (I think) that NetBSD's
OS emulation is broken, and broken in ways that do break real apps;
and second, it may also be hard to hack the scripts such that they
still work on _multiple_ NetBSD hosts using different emuls .

(e.g., so that one script works for both SunOS 4.x binaries on
NetBSD/sparc, and Ultrix binaries on NetBSD/pmax.  If an existing
setup, maybe using NFS-mounts, worked under the relevant vendor OSes,
then why shouldit shouldn't it work under NetBSD's emulation?)

Don't shoot me, I'm just the messenger. As I've said, I see arguments
both ways. And apologies for the repetition.