AT&T in Europe

[Pete Delaney - Rocky Mountain UNIX Consultants]

  I'm working here in France on UNIX for the very hardware it sounds like
the Empire wants to smash; a out of date Honeywell Level-6.  Anyway,
our networking group is working with Oleveite, and, when we propose using
the 4bsd sockets or network support they made a BIG stink about it.  
They claim to want to write their own IPC facility;  however I suspect they
are just following Darth's orders.

	I think the System V COFF format MAY be worth switching to.
Though it is slower (2x?), the ability to place segments, like a
system shared library, may be worth the expense.  Has 4bsd any support
for such things?

	This brings up an interesting subject.  What to do on those
machines where an int is 16 bits to facilate the 4bsd IOCTL's.  The
C grammer says switch(long) is treated as a switch( (int)long ); no?
The concept of copying ioctl data on syscall entry and exit has some
advantages over the copyin/copyout's being at each ioctl. For example
changing the size of a ioctl data structure. The Empire's method will
write on areas outside a data structure if the ioctl struct is extended.

	For now I plan to CAST ioctl cmd's to a int following the ioctl
system call and replace cmd decl's with t_ioctl_cmd typedefs decl's.
The file control stuff seems to do this a little.
I'm Curious what Berkely does on pdp11's.  

	Any X25 to 4bsd TCP/IP interfaces available?  Looks like this
may become a standard here very soon.  

	As far as System V being a standard, well the 3b20 duplex DMERT
is not exactly the same UNIX.  Actually, looks much more comprehensive, 
possibly running networking as a real time process.  System V has no such
facility, 4bsd does.

					Pete Delaney
					Rocky Mnt UNIX
					Honeywell BULL
					France.
					decvax!mcvax!vacnam!lvbull!pete