>> [...sgtty.h...termios.h...]  The old sgtty.h interface is not
>> eight-bit clean.

> Huh?  Did anybody actually go through the trouble to *break* sgtty?
> AFAIK, sgtty has been "8-bit clean" at least since V7.

That was my reaction too...but then I tried it.  We have a mtXinu
4.3+NFS machine still running, and its tty driver appears to be pure
4.3.  In order to get 8-bit-clean output, I had to turn on LITOUT.
Even PASS8 wasn't enough.  (RAW would presumably have worked; I didn't
bother trying that, when I found LITOUT worked.)  I considered ranting
a bit about switching to the SV tty driver instead of fixing the
Berkeley one, but decided ranting on an only marginally supportable
point was probably counterproductive.  (Except for its SV
taint-by-assocation, the termio.h/termios.h interface is IMO a cleaner
interface.  More awkward in some respects - you have to frob half a
dozen things for most common changes like raw mode - but more
orthogonal.)

Thus, while the <sgtty.h> interface is not inherently 8-bit-unclean, it
appears on the basis of this one simple test that a faithful Berkeley
tty driver emulation _is_ 8-bit-unclean.

					der Mouse
					
			    mouse@collatz.mcrcim.mcgill.edu