[comp.unix.microport] System V/AT 286 2.3 upgrade hints, trips and traps

karl@sugar.UUCP (Karl Lehenbauer) (03/05/88)

(This may not be timely for most people upgrading to 2.3; I posted it
a long time ago when the group was moderated - it set for a week at
uwvax and was returned because uwvax couldn't reach uwmacc - the
legacy of moderated newsgroups.  Anyway, we missed the message to
make the group unmoderated -  I got busy - and the rest is history)

Experiences, Hints, Trips and Traps When Upgrading To Microport System V/AT
---------------------------------------------------------------------------

If you don't do the install while running in single user mode, a bunch of 
stuff won't copy because the file being replaced is executing.  I didn't 
see this anywhere in the install doc.  Don't forget to mount your /usr 
filesystem by hand, of course.  (and "they" said Unix was hard to use ;-)

If 'csh' is your root login's default shell, csh won't get copied in.
You'll have to re-edit /etc/passwd to make root use 'sh' instead, or
whatever you do to run root under sh rather than csh.  Again, this is
not in the install doc.

The install docs list some system configuration files (/usr/lib/uucp/L.sys,
/etc/gettydefs, etc.) that the install overwrites, but they don't mention 
files in /usr/spool/cron, at.allow, at.deny, root's crontab, gettydefs or
USERFILE.  (I know I am expected to make a backup first but don't fake me 
out by giving me a partial list of system config files that will be 
overwritten.)  Also, why overwrite a file when nothing about it's format
or anything has changed?  For example, there was no change in the 
L-dialcodes format at all.  The install will overwrite it anyway.

The second runtime disk does a:

	mv: /usr/lib/terminfo/a/ansi.old not found

It doesn't seem to be a problem.  The same cannot be said for the ANSI
driver itself.  Jay Maynard (splut!jay), Steve Nuchia (nuchat!steve)
and I all had problems with vnews after installing the upgrade - big,
weird problems like character pairs occasionaly being reversed and
stuff, tab stops not working - totally gross!  I still don't know what
the problem was, or how it got past quality control (ha ha ha), but
the fix for me was to use the ansi driver from DOSmerge (which,
incidentally, won't work on my machine - but we won't go into that
right now)

Finally, while it's cute to not have changed or even recompiled 'ps' from 
the AT&T 286 distribution (they make a point of mentioning it), it would 
have been nicer still to fix it so that it would automatically use the 
currently running image rather than making us link /unix to whatever image 
we're using.

What's the final score?  I still have the MODEM hang double panic, and
often.  The much-awaited 2.3 release was supposed to fix it.
It has not, at all, but telling those of us having the problem that it 
was going to be fixed in this (late) release certainly got everyone to 
stop phoning, eh?  I'm kind of worn down at this point, on this matter.

Oh yeah, as long as I'm kind of flaming anyway, a "ship date" is not
just when you start shipping - it's when I get the product.  A long time
passes between when Microport starts shipping and when everyone who
has ordered (and paid for :-( the product receives it.

Ta.
-- 
"Lack of skill dictates economy of style." - Joey Ramone
..!uunet!nuchat!sugar!karl, Unix BBS (713) 438-5018

mem@zinn.UUCP (Mark E. Mallett) (03/11/88)

In article <1514@sugar.UUCP>, karl@sugar.UUCP (Karl Lehenbauer) writes:
> 
> It doesn't seem to be a problem.  The same cannot be said for the ANSI
> driver itself.
>  [etc.  stuff about new ansi not working removed -mm]

When I upgraded to 2.3, almost NOTHING that used the ansi console worked
anymore.  vnews, jove, vi, you name it.  I finally hooked up another
computer to my serial port and spent many hours debugging the escape
sequences that came out of these utilities.  I found several problems, and
reported them publicly in the microport conference on BIX (Byte's online
service).  Microport acknowledged these fixes.  Rather, they came back and
said, "look what we found out", and then proceded to reiterate most of what
I'd said.

Unfortunately I've long forgotten what-all was exactly changed.  Much of it
was changing 0-offsets to 1-offsets in things like hpa and vpa.  At any
rate, here is the terminfo entry for ansi that resulted.  (to install it,
save it in a file, say 'foo', and do the command 'tic foo'.  First, you
should make sure you preserve your current definition in case there's a
problem).: 

ansi|ansi-24|ansi-s-bot|ansi-bot-s|ansi with bottom status line (for sysline),
	am, hs, mir, msgr, eslok, xon, cols#80, lines#24, wsl#80, 
	cbt=\E[Z, bel=^G, cr=\r, csr=\E[%i%p1%d;%p2%dr, 
	clear=\E[H\E[J\E[3;2m, el=\E[K, ed=\E[J, hpa=\E[%i%p1%dG, 
	cup=\E[%i%p1%d;%p2%dH, cud1=\n, home=\E[H, civis=\E[25;27m, 
	cub1=\b, cnorm=\E[23;27m, cuf1=\E[C, cuu1=\E[A, dch1=\E[P, 
	dl1=\E[M, dsl=\E7\E[?6l\E[25;1H\E[K\E8, smacs=\E[12m, 
	blink=\E[5;3;1m, bold=\E[1m, smir=\E[4h, invis=\E[8m, rev=\E[7m, 
	smso=\E[7m, smul=\E[4m, ech=\E[%p1%dX, rmacs=\E[10m, sgr0=\E[m, 
	rmir=\E[4l, rmso=\E[m, rmul=\E[m, fsl=\E8, is1=\E[=h, 
	is2=\E7\E[1;24r\E8\E[?6h\E[A\n, il1=\E[L, kbs=\b, kdch1=\E[P, 
	kdl1=\E[M, kcud1=\E[B, kf1=\EOP, kf10=\EOY, kf2=\EOQ, kf3=\EOR, 
	kf4=\EOS, kf5=\EOT, kf6=\EOU, kf7=\EOV, kf8=\EOW, kf9=\EOX, 
	khome=\E[H, kich1=\E[@, kil1=\E[L, kcub1=\E[D, knp=\E[V, 
	kpp=\E[U, kcuf1=\E[C, kcuu1=\E[A, dch=\E[%p1%dP, dl=\E[%p1%dM, 
	cud=\E[%p1%dB, ich=\E[%p1%d@, indn=\E[%p1%dS, il=\E[%p1%dL, 
	cub=\E[%p1%dD, cuf=\E[%p1%dC, rin=\E[%p1%dT, cuu=\E[%p1%dA, 
	pfkey=\E[%p1%d;%p2%l%dq%p2%s, pfx=\E[%p1%d;%p2%l%dq%p2%s, 
	rc=\E8, vpa=\E[%p1%dd, sc=\E7, ind=\ED, ht=\t, 
	tsl=\E7\E[?6l\E[25;%p1%dH\E[7;1;55m\E[K, 



> What's the final score?  I still have the MODEM hang double panic, and
> often.  The much-awaited 2.3 release was supposed to fix it.
> It has not, at all, but telling those of us having the problem that it 
> was going to be fixed in this (late) release certainly got everyone to 
> stop phoning, eh?  I'm kind of worn down at this point, on this matter.

It also got people, like me, to send in money for the upgrade.  In my case,
to sign up for the upgrade policy.  I was told on the phone, "oh that's
fixed in 2.3" and that it was probably a good idea to buy the $99 upgrade
if I was going to get 2.3, since they ship out "3 or 4" upgrades per year.
I wonder how true that is?

-mm-
-- 
Mark E. Mallett  PO Box 4188/ Manchester NH/ 03103 
Bus. Phone: 603 645 5069    Home: 603 424 8129
uucp: mem@zinn.UUCP  (...decvax!elrond!zinn!mem   or   ...sii!zinn!mem)
BIX: mmallett