Path: utzoo!mnetor!uunet!husc6!cmcl2!nrl-cmf!ames!pasteur!ucbvax!cbosgd!ihnp4!ihlpa!kav
From: kav@ihlpa.ATT.COM (Vangness)
Newsgroups: comp.os.minix
Subject: Comments on Jim Paradis' serial tty
Message-ID: <6892@ihlpa.ATT.COM>
Date: 20 Jan 88 00:40:43 GMT
Organization: AT&T Bell Laboratories - Naperville, Illinois
Lines: 58
Keywords: MINIX

-
	Here are some of my experiences with the new console/rs232
	driver package posted by Jim Paradis. First off, I couldn't
	wait to use my serial tty port under MINIX! Thanks Jim!
	The following comments are broken into two sections, the
	console.c section and the rs232.c section.

console
-------
	- moving right in elle causes a destructive space.
	  This may be happening because I am using the
	  original MINIX ansi termcap. I'm not yet using the one
	  Jim Paradis posted with the driver (I want to make that
	  the first file I transfer using the rs232 line!)
	- typeahead before prompt is not killed when @ is used.
		for example:

		# cc foo.c
		ls -l a#.ouf@
		# ls -l
		als not found
		-l not found
		#

		i.e., I started a cc, typed an ls -l before the
		cc finished, the prompt came back during my typeahead,
		I made a mistake and hit @ and only '.ouf' was
		killed - the `ls -l a' part was still in the buffer.

	I'm not sure this is in the console.c portion. It might be
	in the tty.c portion.

	- on an AT&T 6300+ (CGA compatible) the screen gets
	  blanked during a scroll. Any subsequent text doesn't
	  appear. Hitting Ctrl-Alt-F1 a few times straightens it out.

rs232
-----
	- I have hacked/ported a version of kermit to MINIX, but
	  as characters are received, the overhead of servicing the
	  interrupts causes everything else to stop, including
	  kermit. For example, a file of about 24 lines is
	  displayed on the host system, and during the transmission
	  of the characters, only about 2 or 3 characters  are
	  actually displayed by kermit. When the characters are all in,
	  kermit catches up and displays the rest of the screen.
	  I haven't tried tweaking the parameters in the driver that
	  affect how many characters are handled per interrupt, or the
	  latency calcultion.
	  This all happens on an AT&T 6300+ (80286) with communications
	  at 1200 baud. No characters are lost, just the annoying overhead
	  of processing them.


	  
		Kurt Vangsness
		AT&T Bell Laboratories - Naperville, Ill
		ihnp4!ihlpa!kav