Copy Link
Add to Bookmark
Report
Info-Atari16 Digest Vol. 90 Issue 537
=========================================================================
INFO-ATARI16 Digest Sat, 12 May 90 Volume 90 : Issue 537
Today's Topics:
Determining Own Name
Phantom Typist(s) [very long!!]
Re; poolfix3,poolfix4,....
Re~2: Smalltalk ??
Smalltalk-80
Smalltalk 80 price
----------------------------------------------------------------------
Date: 12 May 90 17:15:14 GMT
From: ncs.dnd.ca!balkwill@rutgers.edu (R. J. Balkwill)
Subject: Determining Own Name
Message-ID: <887@ncs.dnd.ca>
I goofed in my previous post on how a program can determine its name.
Missed it by one byte!
The name you want is still at offset 30 from the DTA but its format
is:
name 8 bytes padded with spaces
ext 3 bytes padded with spaces
attr 1 byte (attributes)
null 1 byte
---
Bob
------------------------------
Date: 12 May 90 17:13:15 GMT
From: usc!zaphod.mps.ohio-state.edu!van-bc!ubc-cs!alberta!ccu!bright@ucsd.edu
(Bob Bright)
Subject: Phantom Typist(s) [very long!!]
Message-ID: <1990May12.171315.12165@ccu.umanitoba.ca>
A Poor User's Guide to the Phantom Typist(s)
--------------------------------------------
[This is very long -- about 13K -- so hit 'n' now if you're
not interested. You were warned.]
I'm typing this in Word Perfect, with the phantom typist
screwing around in my machine. I've a few details to report on
the care and feeding of the critter.
(I'd better reboot first; typing with the phantom in tow is
just too painful.)
There, that's better. To begin with, a point of clarifica-
tion. There are TWO distinct phenomena which have been referred
to as the "phantom typist".
The symptoms of the first are that you'll be typing away,
and suddenly the last few characters you typed will start slowly
repeating themselves over and over again on your screen. (I
_think_ it's always been 3 chars in my case, but this may be
accidental.) This bug is quite rare, and quite DEADLY. (I've
seen it maybe 4 times in two years of heavy usage; in fact I'd
forgotten all about it until it struck a couple of days ago --
@#$%~&*!) Once it hits, you're toast. Keyboard and mouse input
have no effect whatever, so there's no opportunity to save your
file; all you can do is sit and watch the same few chars get
repeated until your machine crashes (20 seconds or so, in my
experience). A warm boot makes it go away.
If I'm not mistaken, the term "phantom typist" was original-
ly coined by someone at WPCorp. to refer to this behaviour. I
believe they claimed that it was due to a bug somewhere or other
in TOS, but that they couldn't get Atari Corp. to listen. Let's
call it the "phantom repeater" to avoid confusion.
The second phenomenon that is sometimes referred to as the
phantom typist is the system slowdown that's been much discussed
of late on the net. (Call it the "phantom slowdown"; it may or
may not be related to the phantom repeater.) Typical symptoms
are that you'll be typing away and keyboard/mouse input suddenly
slows to the speed of molasses. No chars actually appear other
than those you type; it's just that your mouseclicks/keypresses
take effect painfully slowly (ranging from numbers of seconds to
minutes). Moving the mouse around makes the keypresses or button
clicks take effect immediately, however. You can usually save
your document and exit normally to the desktop, but the phantom
persists at the desktop level. A warm boot makes it go away.
It was the phantom slowdown that was afflicting my machine
a few minutes ago. It struck not long after I booted WP, and
before exorcising it I decided to do a bit of experimenting.
Here's what happened.
I was typing in WP (my usual speed, not too fast but not too
pokey -- maybe 40 words/min), and everything was fine. I paused
to read what I had written, cursored up a few lines, then down a
few lines (a curious habit that most computer authors seem to
pick up sooner or later), and wham, the phantom had me. Whether
it was there before I hit the up-arrow key, I don't know. I do
know that it didn't show any symptoms before I hit the up-arrow
key, because I watch the screen pretty carefully when I write.
Anyway, I hit F7 to save my work, exited to the desktop, and
decided to experiment a bit before rebooting. Someone had
recently reported that the slowdown appears to affect only GEM
prgs, so the first thing I tried out was some non-GEM stuff.
Sure enough, Moshe Braner's GNOME.TTP was a speedy as ever. DC
SHOWIT also worked fine; I could cursor up and down through text
files with no slowdown at all. (Thanks, Mike; great hack!) Then
I tried DCOPY312.PRG; it's a non-GEM prg which optionally allows
you to select files with the GEM file selector. Keypresses took
effect immediately at the main menu, but as soon the file
selector was invoked the phantom reappeared in all its sticky-
molasses glory. Hmmm.
While I was playing around with the GEM selector in DCOPY I
noticed something that I hadn't before: when the phantom is
active, you can short-circuit the wait between keypresses and
actions not only by moving the mouse, but also by hitting the
Shft, Ctl, or Alt keys. I thought that warranted more investiga-
tion, so I exited DCOPY and ran WP again.
The first thing I noticed was also something that hadn't
occurred to me before: When the phantom is active at the desktop,
auto key repeat works as normal (as evidenced by the keyclicks --
the chars still don't appear in dialogues until you move the
mouse); whereas the phantom disables auto repeat in WP. Auto
repeat also works in GEM accessories invoked from the desktop,
but not when they're invoked from within WP.
OK, now what about the Shft/Ctl/Alt thing? Sure enough,
hitting an alphanumeric key and then pressing (L-or-R) Shft
caused the key to take effect immediately in WP. There was some
weirdness with the Ctl and Alt keys, though. (As it turns out,
this wierdness appears to be due to the way WP maps the key-
board.) What seemed to happen is this:
If you want an alpha char in your document, just press alpha
(doesn't matter whether you release it or not) and then press
Shft; the char will appear on your screen as soon as the Shft key
makes contact. If you want to input a Ctl/Alt-alpha combination,
however, then you press alpha _followed_ _by_ Ctl/Alt. I.e.,
when the phantom is active in WP, alpha followed by Ctl/Alt does
what Ctl/Alt-alpha does under normal circumstances.
This also works for function keys (and this time all of
Shft/Ctl/Alt work the same). If you want shifted F1 (invokes the
super/subscript menu), you press F1 by itself (again, doesn't
matter if you release it or not), and then as soon as one of the
Shft keys makes contact the super/subscript menu will appear.
Similarly, F1 followed by Alt gives you the thesaurus, while F1
followed by Ctl invokes the shell (or rather, it would if WP-ST
allowed access to a shell).
How do you invoke commands bound to _unshifted_ fn keys?
(To get a normal alphanumeric you hit the alphanumeric followed
by Shft, but doing this with fn keys gives you the _shifted_ fn
key command.) Simple: You just hold down either Ctl or Alt
(_not_ one of the Shft keys; that won't do anything at all), then
hit the fn key you want, and then as soon as you _release_
Ctl/Alt the _unshifted_ fn key command will take effect. (That's
simple, right? :-) A similar sequence works for shifted alpha-
numerics: if you want a capital R, e.g., you hold down Shft, then
hit the r key, and as soon as you _release_ Shft the R appears.
(This appears to be about as close to normal behaviour as the
phantom will tolerate in WP.)
As mentioned, some of this weirdness appears to be due to
the way WP maps the keyboard. I had STalker (the terminal
accessory) resident when the phantom struck, so I also played
around with it for a bit. Pressing any alphanumeric followed by
any of Shft, Ctl, or Alt caused the char to be immediately sent
to the modem. STalker also uses Shft/Ctl/Alt combinations for
various commands; to get them when the phantom is active you hold
down Shft/Ctl/Alt, then hit the appropriate alpha or function
key, and when you release Shft/Ctl/Alt the command takes effect
(just like the shifted alphanumerics in WP).
Incidentally, hitting any of Shft/Ctl/Alt also makes chars
appear immediately in desktop dialogues.
To sum up so far (and generalizing somewhat):
1. The phantom slowdown appears to afflict only GEM windows and
dialogues (for more on this point, see below). When it's active,
keypresses can be forced to take effect immediately by changing
the state of one of the Shft/Ctl/Alt keys. Commands that would
normally be invoked with a Shft/Ctl/Alt combination can be
invoked when the phantom is present either by pressing the
appropriate key first and then the Shft/Ctl/Alt key (WP), or by
holding down Shft/Ctl/Alt, pressing the appropriate key, and then
releasing the Shft/Ctl/Alt key (STalker, upper case in WP).
A few other interesting facts I ran across in my experimen-
tation:
2. When the phantom is active, multiple key presses are buf-
fered, but only up to exactly 9 chars. (?!) I.e., if I hit <9
alphanumerics in a row, and then pressed Shft, all of the chars
would appear immediately. But if I hit >9 alphanumerics before
pressing Shft, exactly the first 9 appeared on the screen; the
rest were lost forever. Caveat: The foregoing applies to both WP
and to desktop dialogues. In STalker, however, while things
worked similarly, exactly _8_ chars were buffered instead of 9
(??!!; beats hell out of me).
3. STalker supports fn key macros which can be up to 40 chars in
length. Interestingly, the 8-char buffering limit doesn't apply
to these macros. I.e., if I hit F1, and then hit Shft to
temporarily short-circuit the phantom, all 40 of the chars in the
F1 macro would immediately appear on the screen.
4. This one virtually guarantees that the phantom resides
somewhere "higher up" in the OS, and that it's not (at least not
exclusively) a problem with the IKBD. I use a corner clock
called JCLOCK7 (by John Stanley, I think; thanks John, if you
happen to read this). JCLOCK7 toggles its am/pm indicator from
lower to upper case to indicate the status of caps lock. WP also
has its own caps lock indicator in the status line at the bottom
of the active window. When the phantom was active in WP and I
hit the caps lock key, the JCLOCK7 indicator would invariably
update at once. The WP indicator would update, however, only
when I moved the mouse or pushed one of the Shft/Ctl/Alt keys.
I.e., the caps lock key was being read and processed immediately,
but somehow or other the phantom was blocking the transmission of
its status to GEM. Moving the mouse or changing the state of
Shft/Ctl/Alt temporarily unblocks the transmission, at least for
the last nine keyboard events.
5. I'm not sure what relevance this one has (I've included it
mainly because I'm going to send a copy of this note to the ST
programmer at WPCorp., and it may give him a better idea of where
the phantom actually resides). When you hit F10 in WP ("Save"),
the program puts up the GEM file selector so that you can select
the name of the file to save. When you hit 'OK' the GEM selector
disappears, then WP checks the disk directory to see if the file
you specified already exists, and if so, asks whether you want to
replace it.
That's what should normally happen. When the phantom is
active, however, and you invoke the Save command (which again,
you can do without moving the mouse only by holding down Ctl/Alt
before hitting F10), things work a bit differently. You press
Ctl-F10, and when the Ctl key is released the GEM selector pops
up. Hit Return to use the default filename. Nothing happens (as
expected; the phantom is blocking transmission of the Return, and
we haven't done anything to unblock it yet). Now hit Shft. The
GEM selector immediately disappears, but then....nothing. What
_should_ happen at this point is that WP should check the direc-
tory to see if the specified file exists, and then put up the
"Replace?" prompt. But it's not doing anything at all (I'm quite
certain of this, because I removed the disk and then reinserted
it before I hit Return, in order to insure that the directory
wasn't cached). It's just sitting there.
If you now press Shft a second time, however, the directory
gets read at once and then the "Replace?" prompt comes up. In
other words, in this case we need to unblock the phantom TWICE,
with two separate presses of Shft. The first one unblocks the
Return, while the second one unblocks -- well, who knows what?
I guess that's all I have for now. I have no idea whether
the phantom slowdown is due to an OS bug, or a bug in WP, or a
bug in one of the multitude of TSR's and accessories I normally
use. I know for a fact that some of the latter wouldn't pass
muster as "well-behaved" (JCLOCK7, e.g., won't run under TOS 1.2
(I'm using 1.0), so it must be doing something naughty). On the
other hand, the sheer number of people that have reported the
problem in a variety of contexts makes it a priori unlikely that
it's due to a bug in any single program, suggesting that it's due
either to an OS bug or to a very common applications programming
mistake.
But on yet another hand (doesn't everybody have three? :-),
there's the fact that some users have recently reported the
phantom appearing under Spectre; which doesn't jibe very well
with the fact that its characteristic behaviour in native ST mode
seems to be to block transmission of keyboard events to GEM.
(Dave Small, are you out there?).
What really baffles me is that all of the stuff that I
normally like to run can co-exist very happily together for days
or weeks on end; and then all of a sudden, wham, the phantom
appears out of nowhere. Anyway, as mentioned, I'm going to send
a copy of this to WPCorp. in hopes that it will give them a few
ideas on where to look. In the meantime, let's keep trying to
narrow down the hypotheses.
[And to the good folks at Atari: You've been awfully silent
about the phantom since kbad suggested the "chicken blood on the
keyboard" fix. Any and all suggestions, hints, advice, whatever
would be most appreciated. Have any of you experienced the
phantom slowdown yet? What do you think I should try the next
time it strikes?
[Maybe you could suggest to the boys "upstairs" that they
offer a reward to the first person that isolates the problem and
figures out a fix -- a couple of free TTs, say, when they become
available? Seems to me it would be well worth it. Whether the
phantom is an Atari bug or someone else's, the person who finally
tracks it down is someone you definitely want developing on your
new machines.]
BBB
--
Bob Bright <bright@ccu.umanitoba.ca>
Dept. of Philosophy
University of Manitoba
Winnipeg, Man R3T 2N2 (204) 474-9680
------------------------------
Date: 10 May 90 09:09:12 GMT
From:
swrinde!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!ira.uka.de!smurf!gopnbg!
mcshh!tpki!feki!marten@ucsd.edu (Marten Feldtmann)
Subject: Re; poolfix3,poolfix4,....
Message-ID: <a911428@feki>
>From: csbrod@medusa.informatik.uni-erlangen.de (Claus Brod )
>Date: 7 May 90 16:32:43 GMT
>
>Dear Mr Pratt,
>
>since I don''''t have regular access to Usenet, this letter might be somewhat
>late. People have told me about your reactions towards my version of POOLFIX,
>and I would like to add some comments.
>[.....]
Good answer -- a lot of ST user here in Germany think the same way.
Well done, Claus!
Marten
M. Feldtmann, Eckernfoerder Str.83, 2300 Kiel 1, West-Germany
SUBNET:marten@feki DNET\EUNET\USENET:marten@tpki.de
------------------------------
Date: 10 May 90 20:19:35 GMT
From:
swrinde!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!ira.uka.de!smurf!gopnbg!
mcshh!schiers@ucsd.edu (Carsten Schiers)
Subject: Re~2: Smalltalk ??
Message-ID: <7467@mcshh.hanse.de>
stefan@hpbbi4.HP.COM (#Stefan Bachert) writes:
>> The normal distribution is about DM 400,-.
>This is NOT the normal distribution. This is a special price
>for students. The NORMAL price is about DM 2400.
For the special price you get the normal distribution. Everthing clear
now? It was meant to be aware of the fact that nothing is missing.
> Add DM 200 for books, the manual is not valuable for
> learning smalltalk
Indeed. This also proves for GNU-Smalltalk.
And now, I have to tell you how nice I find ParcPlace
Smalltalk-80, since otherwise I have included more
than I tell you now. And then, the mailing system
would reject this a third time.
I think this is crazy, since I had to include that much
to give the context.
CU Carsten.
unido!imdm.uke.uni-hamburg.dbp.de!schiers
unido!netmbx!mcshh!schiers
------------------------------
Date: 10 May 90 08:49:48 GMT
From:
swrinde!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!ira.uka.de!smurf!gopnbg!
mcshh!tpki!feki!marten@ucsd.edu (Marten Feldtmann)
Subject: Smalltalk-80
Message-ID: <a129462@feki>
>From: buggs@cup.portal.com (William Edward JuneJr)
>Date: 6 May 90 22:28:17 GMT
>>CU Carsten.
>>Smalltalk-80 from ParcPlaceSystems at Georg Heeg at Unido.
>>I do not know, whether he is allowed to distribute it outside Germany.
You see,Atari ST is quite good used in Germany! Atari-US really should look
at TURBO-C to produce the TOS. With the release 2.0 of this compiler, one
can produce absolute-code!
With ST-80: you just can get version 2.3 from them. Objectworks is not
supported and will not be in the future (so said at the CEBIT), because
it needs so much memory, that even a Mega-ST4 will not satisfy it. Never
version will be supported, when ParcPlace releases better Memory-Manager.
Perhaps they will port Objectworks to the TT...
V2.3 works with the big screens from Matrix and (sometimes with a little patch)
from Atari. The performance of the implementation:
Standard-ST-Benchmarks: 40-42
with Hypercache (16 MHZ Accelerator): 70
Quite good for this lovely machine. With a 386-machine, 20 MHZ, I could get
with Objectworks V2.5 just an Index of 102.
Marten
M. Feldtmann, Eckernfoerder Str.83, 2300 Kiel 1, West-Germany
SUBNET:marten@feki DNET\EUNET\USENET:marten@tpki.de
------------------------------
Date: 10 May 90 20:11:45 GMT
From:
swrinde!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!ira.uka.de!smurf!gopnbg!
mcshh!schiers@ucsd.edu (Carsten Schiers)
Subject: Smalltalk 80 price
Message-ID: <7466@mcshh.hanse.de>
stefan@hpbbi4.HP.COM (#Stefan Bachert) writes:
>> The normal distribution is about DM 400,-.
> ~~~~~~
>A friend of mine bought smalltalk a year ago. And if things didnot
>change the above statement isn't quite correct.
>This is NOT the normal distribution. This is a special price
>for students. The NORMAL price is about DM 2400.
For the special price you get the normal distribution. Everthing clear
now? It was meant to be aware of the fact that nothing is missing.
>PS:
> Add DM 200 for books, the manual is not valuable for
> learning smalltalk
Indeed. This also proves for GNU-Smalltalk.
CU Carsten.
unido!imdm.uke.uni-hamburg.dbp.de!schiers
unido!netmbx!mcshh!schiers
------------------------------
End of INFO-ATARI16 Digest V90 Issue #537
*****************************************
