In some message, a strange person (i.e. me) writes: 
>I'd have to disagree that sunos emulation is somehow easier under netbsd 
>than solaris. The degree of code commonality is if anything higher 
>between sunos and solaris than between sunos and netbsd. 

OK, how can you say such a ridiculous thing, people are asking me? 
Basically, what am I saying is if you have written kernel code for sunos 
4.1.3 that is:
1) a VFS
2) uses the VM subsystem to do lots of mucking with pages, segments, or 
   process address spaces
3) uses modloadable .o's for system calls or vfs
4) does Kernel RPC

Then you will probably find, as I have, that moving said code from sunos
4.1.3 to solaris is easier than moving it from 4.1.3 to netbsd--
***provided you have source to solaris ***. In fact, the only hard part of
the transition for kernel code from 4.1.3 to solaris is device drivers,
due to the DDI stuff. But strangely enough I've found the task of moving
kernel code to solaris to be far *easier* than the task of moving user
code from sunos to solaris. 

The hardest parts all involve solaris networking (a mess in kernel or user
land). 

So, after looking into solaris innards, and seeing what Frame needed to
run in compatibility mode, I could find no real *technical* reason that
frame would not run under solaris 2.0 through 2.2. The missing pieces were
pretty simple, and should have been there, and evidently are there now. 

Hence my comment re the three OSes. User-land is another story, of course ...

ron



------------------------------------------------------------------------------