Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 UW 5/3/83; site uw-beaver
Path: utzoo!watmath!clyde!burl!ulysses!mhuxj!houxm!hogpc!houti!ariel!vax135!
cornell!uw-beaver!info-mac
From: info-...@uw-beaver.UUCP
Newsgroups: fa.info-mac
Subject: experience at Rutgers with large intro courses
Message-ID: <2195@uw-beaver>
Date: Mon, 12-Nov-84 03:46:01 EST
Article-I.D.: uw-beave.2195
Posted: Mon Nov 12 03:46:01 1984
Date-Received: Tue, 13-Nov-84 19:19:51 EST
Sender: yenbut@uw-beave
Organization: U of Washington Computer Science
Lines: 205

From: Charles Hedrick <HEDR...@RUTGERS.ARPA>
This message is an attempt to summarize our experience this fall.
We are using Macs and Lisas for our first three computer science
courses.  These include an intro programming course, a second semester
of programming (emphasizing data structures), and a course in
assembly language and machine architecture (emphasizing the latter -
we don't consider assembly language per se to be worth all that much).
We have around 60 Macs, used for the first two courses, with MacPascal.
We have around 30 Lisas 2/5's, used for the third course, with System V.
I am not going to make detailed comments on the software.  Our faculty
are in a better position to do that.  I can comment mostly on the
general logistics of handling over 1000 students.  Our machines
are distributed over 5 locations.  All but one are attended, and are
open until around 1am.  The person attending them has other jobs,
however, like distributing printout, and handling questions about
our DEC-20's and other machines.  The largest site is at the computer
center, and is open all the time.  It is not attended.  We have
secured the machines using glue-on pads.  (They come in two pieces,
which glue to the bottom of the machines and the table.  They are
held together by rods which are held in with a lock.)  So far we
have not lost any machines, but have lost 2 mice at the unattended
site.  (We generally do not have much of a security problem.  We
normally leave terminals in open rooms without securing them, and
have had very little trouble.)  

Our major hardware problem is with diskettes getting "eaten".  Several
of our Lisas eat diskettes.  That is, suddenly your files are not
readable.  Unfortunately, our students are not always able to determine
which machine is causing the trouble, except in extreme cases.  So we
don't know the exact number that are causing trouble.  It looks like
around 4.  One of our biggest headaches is the fact that there is no way
to check alignment of a drive.  We are trying to get Apple to send in a
level 2 service person to do this.  The claim is that it will take 5
hours per drive.  It is hard to believe that we can't figure out which
ones are bad, but we really can't.  By the time the damage is noticed,
it is not clear where it was caused.  We seem to have the same problem
with the Macs.  One of them (the first one we tried) had this in a
fairly reproducible form.  After using Pascal in it for a few days, the
disk would become unbootable.  Of the public Macs, we are quite sure
that we have one bad drive (though we don't know which), and suspect
that there are a few others.  So somewhere  between 2 and 6 out of 60
seem to have been bad.  Otherwise, the hardware has been good.  We have
had a Profile failure or two, and a couple of other infant mortalities
on the Lisas, but the Macs have not had any other problems that I can
recall.  I was concerned that the Imagewriters would not stand up to
student use. We have two at each site, one for the Lisas and one for the
Macs. (total of 6).  We have had one failure so far, which isn't bad
considering our past experience with micro printers.

Now we come to the logistics.  We were unable to get copies of MacPascal for
each student.  There are two problems here.  One is pricing.  We don't think a
public institution can require students to pay > $100 for software for a
required course.  Apple agrees, and we expect to see a special deal for use in
courses.  However there have been infinite delays in getting those
arrangements made. Even now, although apparently everybody agrees on the
terms, there is no final program.  The best suggestion so far is that they
will sell the regular package for $75, and we can distribute the 2 system
disks to 2 separate students, for a cost of $37.50.  We think that is still a
bit too high, but there are also some slight problems, like the fact that
there is only one set of manuals, and the fact that we don't want to be
responsible for any bad diskettes that might turn up.  (Apparently the
students would have to come to us if they had problems, and it would not be
clear that we could get Apple to replace the diskettes.)  But when we tried to
do that, it turned out that they simply had not made enough Pascal diskettes,
and could not do so in time to be of any use.  For this fall, Apple has loaned
us one copy of Pascal per machine (plus some extras in case of damage).  We
then have the students check them out, leaving behind I.D.  cards.  This is
fine at the sites that have humans there all the time, but for the one that is
open 24 hours a day, we have had to hire a new set of students to handle this.
And students still don't always return them on time.  It has been an infinite
hassle, and has the students, the faculty, and two different pieces of the
computer center yelling at each other.  (The students don't know this, but it
also turns out that the University will not authorize us to do anything about
students who abuse the system.)  We have been assured that for the Spring,
there will be packages that each student can buy.  We certainly hope so,
because we won't survive this arrangement again.  We would like to put Pascal
on the Sunol network (the only network we have heard of so far).  Everyone at
Apple and Think is very interested and helpful, but among themselves they
can't get anywhere with a licensing agreement to allow this. We suspect that
this problem will not be solved until a year or so after Apple themselves has
a network.  We would be very interested in hearing from other schools that
want to put their software on a network.  Apple claims that they have heard
little demand for this, and none of it representing people who need it
immediately, as we do.  We regard this as the only practical way to give
students access to the software, and will do it within a few days of when it
becomes technically and legally feasiable.

I have a general comment on Mac Pascal.  The Mac user interface is very
sexy.  It is not so clear that students find it easy to use.  From
watching students, I would say that they find it a bit more confusing
than Tops-20.  (These are students who have not seen Tops-20 before.)
And much as I hate to say it, from past experience at U. of Illinois, I
would say that students generally find the old Call-OS or CDC NOS
interfaces easier to pick up than Tops-20.  (For those who don't
remember Call-OS: it was a time sharing system that looked like a
typical BASIC top-level.  OLD, NEW, SAVE, and RUN, and you retyped a
line to change it.)  Our situation is complicated by the fact that
students are sharing diskettes.  It is amazing how many things you can
do to a MacPacal diskette.  They change the background (and all other
preferences options), remove Pascal from the desktop (doesn't do any
harm, but the next student may not be able to find it - and of course
once you have removed it, copy protection won't let you move it back
on in the obvious way.  a really wierd feature.), and even edit the
print program so it does slightly strange things.  (It is supplied in
source form.)  We can't write protect the diskettes because there is a
bug in MacPascal that prevents us from keeping all the students'
programs on a separate diskette.  Thus they have to copy them to the
Pascal diskette, play with them, and then copy them back.  (The bug is
that Pascal doesn't always check whether the right diskette is in the
drive, so it sometimes blows up if you try to run when you have the
student's diskette in the drive.)  Certainly these problems will go away
next semester, assuming that the students do each have their own copy.

The students all hate System V.  First, you should understand that
Rutgers hasn't used Unix in the past.  After Tops-20, it is a great
shock.  So there is a bit of bias against Unix here.  But that isn't
all.  The biggest problem is the Profiles.  We bought them because (like
most universities) money is limited.  They have enough capacity for our
purposes.  But they are ssssllllloooowwww. An ls command seems to take
forever, most of it time for the program to load.  We do not see any
performance difference between loading programs from hard disk vs.
floppy.  In fact by any objective standard, the time to do a command is
perfectly reasonable (about 7 sec. to do an ls).  But is is slow enough
to make the system feel painful.  We have also had problems that are
just learning-curve things.  E.g. we didn't tell the students that they
should save any of their ac's that had useful information before calling
printf. We didn't say it partly because I figured the faculty would
mention in class that you normally save AC's when calling library
routines that don't specifically say otherwise.  Also, partly because we
didn't have time to try things out very much before the semester
started. And our graphics routines don't work.  It seems that in the
first week of class we found that system V was mysteriously crashing on
machines that had printers.  We got a new release of the kernel that
fixed this.  What they didn't tell us was that this release changed the
address to which the screen was mapped.  On a timesharing system this
would be trivial.  But we had 30 machines at 4 locations, each with
their own copy of the software.  This sort of thing will be OK next
semester, because we will know more about the sytsem and have more time
to prepare.  This will still leave us with the "#$'&%  slow Profiles,
though.  The only thing I can think of to do there is use RAM disk.  As
we have no source nor internal documentation, we aren't going to be able
to, but if anyone hears of a RAM disk for the Uniplus System V, I'd love
to hear it.  (PS: one of our staff has a Lisa with a Corvus disk.  Aside
from the fact that the Corvus disk has failed a few times, he loves it.
The problem is strictly with the slowness of the Profile.)

Apple's support has been of varying quality.  I am sure they believe
that they moved heaven and earth to get us the hardware and software in
time for class to start.  And in fact they did.  Just barely. However we
ordered it in late Spring.  The delay was because of attempts to
negotiate a University-wide discount program.  I won't bore you with the
details, beaause it is not clear to me whether it was Apple or Rutgers
slowing things down.  But the evidence does suggest that you shouldn't
hold your breath while negotiating any sort of unusual contract with
Apple.  The problem is that except for that one initial shipment,
everything else has required us to call our salesman several times, and
her to call the regional support center several times.  We still don't
have printer ribbons.  We have been buying them locally, 24 hours before
each assignment. (We didn't do them in advance, because we were told
that the ones from Apple were in the mail.)  We now hear that they are
on backorder (and Apple suspects that the next time we try to get them
locally we will fail there too).  We still don't have a copy of Inside
Mac.  It's really great to hear all the neat stuff the rest of you have
been doing. We wish we could do development too.  We were just notified
that there had been a price decrease in Inside Mac.  Our origianl order
(made this summer) has been cancelled so we can order again at the lower
price.  Gee, thanks, guys.  One of our grad students has just become a
certified developer.  We are hoping that this will give us access to
Inside Mac.  We find that Apple hasn't yet figured out that mice may
disappear.  We would like to keep a few spares, so that we can have the
operators swap out ones that need to be cleaned, and replace ones that
walk away.  We can't order them, because only level 1 service centers
can order them.  It now appears that we can get a local dealer to front
for us in replacing stolen mice.  But it appears we still can't keep
spares on site.  I guess we are going to become a level 1 center, but
the whole point of all this was supposed to be to make it easy on the
customer, not to require him to turn himself into a dealer.  We are also
frustrated by not being able to get diagnostics.  We don't want to fix
anything.  But we do want to be able to figure out which Macs have
misaligned disk drives.  However now it seems that the dealers can't
check alignment either, so I guess we shouldn't feel discriminated
against.  We think that Apple's policies about not giving customers
access to support resources are a bad idea.  They force the 
sophisticated customers to become level 1 authorized.  This does not
serve to protect the dealers.  We don't want to replace our local
dealer, but we may be forced to.

I'm sure these problems will all be cleared up within 6 months or so.
But the result has been a very unpleasant fall.  In case anyone wonders
why we did this, I think we didn't have a lot of choice.  We were using
a similar number of Terak micros.  They were failing more and more
often.  Spares were no longer available on some parts.  And our
technician quit.  It seems clear that we would not have survived the
fall using the Teraks.  Our specs required a student-oriented Pascal, a
machine with a decent instruction set, and a symbolic debugger. The Mac
was the only thing (other than another Terak, and Terak didn't respond
to our RFQ) that had an appropriate Pascal.  Unix on a 68000 was the
least expensive thing we could find that met the specs for the assembly
language. (Our faculty do not consider the 8088 instruction set
acceptable.)  So we felt that this was a reasonable way to go.  I'm not
sure what we would have done if we had realized how it was going to turn
out.  Maybe we would have done the same thing anyway.


-------