-----BEGIN PGP SIGNED MESSAGE-----

On Thu, 4 Feb 1999, Ken Nakata wrote:

>> OK, while we're on the subject:  How much work would it take to let
>> compat_svr4 support A/UX binaries on NetBSD, at least to some extent?
>
>Wasn't A/UX based on SVR2 or 3 instead of SVR4?  How much difference
>is there between SVRs 2, 3 and 4?

IIRC, SVR2, with their own work to bring it up to SVR3 SVID compliance,
plus the BSD Net/2 networking code and clients.  It was a hybrid system,
with a system call to switch between Berkeley, SVR2, or POSIX emulations
- -- sort of like our emulations, but switchable on the fly.  As an added
level of weirdness, they included MacOS compatibility -- in the
userland. In other words, you could install MacOS and run it as just
another application.  If it crashed, you hit command-control-e, and were
punted back to the shell.  Even cooler, they shipped it with a modified
version of MacOS 7.01 which could use the Unix TCP/IP stack, and allowed
for `mixed-mode' applications, which ran as a Unix process, but had a
MacOS user interface.  All in all a very odd beast...

Assuming we left unimplemented their emulation switching and other
oddball APIs, my real question is: how different were the SVR2 and SVR4
ABIs for m68k?

- -- 
				Jim Wise
				jwise@unicast.com



-----BEGIN PGP SIGNATURE-----
Version: PGPfreeware 5.0i for non-commercial use
Charset: noconv

iQEVAwUBNrkYL4kLDoBfn5jPAQGPBggAglW42LVzR6Fyt3m9abLoWRFs828Hn3GJ
fvBQr8GhQmDrKVxfqvoLqyCqOdSmONiuIkKunUXk+DCdDBAeNbgZXO6tkNtcr6xm
moBP4qyBrYCy7bdbF6Oz4zOpwRolDfe0BL2fPNfaYUumGTfHjmG9O9wKvowKUYgI
eyvFMJhbbK/cnLqAsVORa1G5QdJPDqHpI70zYIoDt6lbS5DzB9rga7+jdd9dsi6I
wT1FYCpxd3GCk7A6fnLEFkhardqh5bsXxUt5DqtaVm40VYAvThlBwvjapzTVthvL
gTuZASbL6O6jbAhKXuAE9NG/71tJ7L6SKo00fu0nMEkJtLNDUUlkzQ==
=oYSB
-----END PGP SIGNATURE-----