Marketing Unix - (nf)

[Guy Harris]

> They got it right years ago, eh?  How about that brain-damaged
> Berkeley TTY driver--when will they get it right? (BTL's is much
> better...)  They've both got warts, and both have different
> applications.

Well, I think the USG tty driver has a better "ioctl" interface (although
some may disagree), but the Berkeley one has a better user interface (i.e.,
it actually knows how to implement "crt rubout").  The Berkeley one gets
a lot of its cruftiness from a BTL tty driver, namely the V7 driver, which
was a bag on the side of the V6 driver.  The Berkeley driver was a bag on
the side of the V7 driver, but this was done for backward compatibility
(as was the V7 driver, probably).  Some people have complained about
"gratuitous incompatibilities" introduced by the USG driver (although I
have seen a mail message from one of those people referring to the USG
driver's "much cleaner than v7" tty driver interface).

Some of the problems might be cleared up by getting rid of the UNIX 2.0
backward compatibility stuff in the USG driver, and putting in V7 backward
compatibility stuff instead (within AT&T, V7 backward compatibility was
probably less useful than UNIX 2.0 backward compatibility; outside AT&T,
nobody had UNIX 2.0 so V7 compatibility is more useful).  We put that into
our System III system, and it was reasonably happy running a version of
Rogue and a version of "mille" built for a system with a V7 driver.

(The 4.1c driver made an attempt at cleaning up the interface somewhat, by
adding "ioctl"s to get/set all 32 mode bits, and to get/set all the control
characters; this was backed out of 4.2.  The USG driver's interface may have
been considered at some point for the Berkeley driver, but rejected for
reasons of backward compatibility.)

	Guy Harris
	{seismo,ihnp4,allegra}!rlgvax!guy