thanx..

Well, I’m not the best in thw world.. In fact I suck.. It was really
a guy who
doesn’t know a lot, than forget it.. I just wanted it to figure out how to
drive a arrow a little faster than it goes in that DeskTop coco program that
apperared a little while ago in Rainbow.. I’ll just stick with learning the
basics of os9 and see what I can do from there rather than fool with
assembly.. But thanx for your interest.. By the way, really like your Bash
ay, I
shouldn’t complain, I can’t do it!! Is your new Coco 3 Z’89 a good, no think
trigger happy arcade game?

Mike

I’m not familiar with assemblers besides EDTASM (ROMpak & disk versions), MAS
from Tom Mix, the OS-9 assemblers and (the best one) MACRO-80C from the Micro
Works.

The assembler programs in MACRO-80C can be patched to use the 80-column screen
since they’re BASIC programs, (just alter R$) but the editor is sorta linked
to a 32-column screen. No big deal, since very few people USE the editor from
that package, preferring to use TW or VIPwriter instead.

Keep checking the Radio Shack stores… there’s got to be a copy of Disk
EDTASM there somewhere!

Don

Mike, I had written a small program a long time ago which is
basically a “SCAN” program (like A-DOS does) and this program also
adds the proper number of spaces to each $09 character to display
EDTASM+ files properly. By doing a OPEN”O”,#1,”OUTFILE”:POKE
111,1:SCAN”INPFILE”:CLOSE this program will open a text file, SCAN the EDTASM+
source file and
send the tabbed version to the open disk file. This allows converting or
concantenating several source files into one long ASCII text file. There
is no memory size limit and it is (sorry Mike!) faster than EDTCVT.

If you are interested (or anyone else, for that matter) let me know.

….Roger

This month’s issue of The Rainbow (August 88), has an interesting
story about ML graphics. All programming is done is ML. There is
one listing which is a subroutine library, and four sample programs
, each of which uses that same library. I’ve keyed in the SR library
and the first sample program, and get a successfull assembly with no
errors. It won’t function, though. The author states that “You will
need to resave the binary file file with the addresses shown in Table
1.” Table 1 lists start, end, and beginning exec address for each of
the four programs. How do I do this “resave” function?

Mike,

Today I downloaded your EDTCVT.BIN program in the Assembler
Source Data Lib. I am using Disk EDTASM+ V01.01.00. When I try
to convert a program which it wrote out with the WD command,
your program tells me “Not a valid file type”. What am I doing
wrong?

Zack

Radio Shack is now closing out the Disk Editor Assembler (EDTASM +)
for the Color Computer. This is catalo item number 26-3254.
The price is $5.80. Check your local stores for a copy.
If they want more, tell them to look up the price on
the current “spiff” list.

The store near me has FIVE of them in stock presently.
If you live in an area where you simply CAN’T find
a copy of this item, and still want it, and you can’t
bulley your RS store manager into locating one in some
other area because the item is such a tiny sale,
then reach me here thru MAIL to arrange for me
to buy for you one of the copies at the store
near me. HOWEVER… it will cost you $14, to cover my
hassel in going out and getting the item, paying sales tax,
packaging it up, paying shipping costs, and taking it down
to the post office to ship it. $15 for Canadians, for whom
the shipping cost is a hair more.

Note that I STRONGLY urge you to find copies for yourself,
and to use me ONLY as a last resort. The almost three fold
higher price it will cost you if thru me should also be
an incentive for you to shine me on in this offer. But I am
willing to be a “source of desperation”.

—marty

Does anybody know why when I type in an
assembly language program in with the assembler it always starts at about 4F00
and not where it is supposed to according to the examples I am typing I (i.e.
the sample programs in the july 88 rainbow or
an program in the book Assembly Language Programming RS (62-2077) ).
I am using the disk Edtasm program by RS (26-3254) modified with the patch
program PATCHED.BAS in the data library.

Any help would by appreciated. I am trying to learn assembler and this
problem makes no sense to me?

Thanks
BREWINGTON

Steve,

When using the switch between ROMs and RAMs at $FFDE on the CoCo 3
(assuming that the GIME ROM map bits have been set to
map the ROM as it is done on the CoCo 2 with 16K internal
and 16K external), is it true that no matter HOW you’ve set
up the GIME MMU registers, when you switch to ROM/RAM
mode by hitting on $FFDE, the ROM will be mapped into
the upper part of memory? Eh?

Thanks for refreshing my memory on that one.

—marty

Calling all ML programmers!!! Can anyone interpet the following eight byte
ML program for me? I know what it’s supposed to do but I don’t know how to
change
it to fit a different basic program. It is POKEd into memory by as Basic program
I have and it is used to provide a return point from another program. The eight
(hex) bytes are:

45 58 05 52 41 44 41 52

They are POKEd into memory locations 226 to 22D. How do I put this ML program
into a different Basic program and specify an appropriate entry point for it?

THANK you for the information. Also what does the (NR) next to my name mean.
I joined the sig I think.
Jack

I would like to learn how to program
in assembly language. However, I understand that
Edtasm + is not available. Can anyone an assembler and where to get it
and aprox. cost. Thank You
Jack

I’ve heard generally GOOD things about Cer Comp’s BASIC complier.
Indeed, the only consistant “bad” thing I have heard about it
is lots of indignation over the fact that it is copy protected.
But, apart from that, all reports I have heard (including
some from rather respected programmers) is that it
is a fine Basic Complier, and extremely complete.
It apparantly reflects years of updates to make
it extremely compatible with interpreted Disk Basic.

I have not used it myself.

—marty

Any CoCo assembler experts out there?
Some I know is writing a term program using Edtasm, has the hires
screen stuff and bare-bones term routines working great, except(!)
when I press a key and there is stuff coming in. It drops a character
at 1200, and a few of them at 2400. I’m using the POLCAT rom call at
$A000 (indirect), and it must be pretty slow. So, I have to figure
out how to run the RS-232 pak off an interrupt. The specs in the pak
manual are a little buggy, and no example of how to handle IRQ’s is
given in either the pak manual or the Edtasm manual. Can anyone help
me out? The program will be really cool when I get this problem out
of the way, and I can’t wait to make it available. References to any
good books/articles would be appreciated too..
peace, -:=<

JUST WONDERING IF THERE IS SUCH AN ANIMAL THAT CAN TAKE BASIC AND CONVERT IT
INTO SOURCE CODE, AND IF NOT WHY NOTF SO WHAT? THANX
…….KEN

How do you set the palettes in ML..I’ve tried storing a value in FFB0-F and the
computer just hangs up. thanks for your help.
– —Ezra

DID YOU EVER CONTACT COMPUTIZE ? CAN YOU GIVE/SELL ME
THE COLOR MAX CGP-220 DUMP SOURCE?

ANDY.

Hi! Thanks to your relatively lucid explanation of multitasking and such,
you have been selected (unanimously, in fact) to get this little question.
Congratulations!
No, seriously, if you can help me out with this, I would appreciate it
greatly. I am messing with a VERY (x 1e06) small “time-slicer” for my CoCo 2
just out of curiousity as to how such as system works. What I am wondering
concerns the system’s clock interrupt. I seem to vaguely remember that there
is an interrupt every 16.66667 ms or so from somewhere in Mr. Computer. Now,
if this is so, this seems to work out to something like 15000 6809 cycles
per interrupt. Is this about right? And if this interrupt is as described
above, which one is it? Which one of the interrupt vectors at FFFx would I
replace to intercept this particular interrupt? Also, am I right in
assuming that the interrupts of this sort are usually disabled on start-up,
for normal usage, so that I would have to enable the interrupts?
I would be quite willing to find the info myself, but I can’t find it
anywhere. I’ve checked all the sources I have, and none tell me what I need
to know. Can you help? If not, can you refer me to a likely party? Thanx!

Ryan Waldron

Mike,

When writing a memory size test for the CoCo 3,
be VERY careful to note that
in a 128 K CoCo 3 memory GHOSTS to non existant
segments. So testing for 128K vs 512K requires
you store values in the first byte of each of
several memory pages… like address $00000 and $20000…
and then make sure the byte in each segment is different!

—marty

Don,
Say, do you know how to detect a key?
Using basic’s CHROUT routine? I can’t find it in any of my books.

-Mike-

How can you determine if the shift key is being pressed? Under
machine language that is. I need to detect a keys.
Can you help?
-Mike-

I have a question for you. Is there
an addressing mode or a way to use
the index registers like X or Y
for example to hold a end of buffer
pointer?

EXAMPLE:

00010 LDX BUFFER,PCR
00020 LDY BUFFER,PCR
00030 LEAY 256,Y
00040 LDA ,X
-
-
-
-
-
00145 CMPX ,Y *Here’s my prob.

When the CMPX ,Y instruction is executed, the cpu looks for the address
contained it register Y. Not the value or buffer pointer that Y holds.

I am tstill learning about 6809 programming. I used to program in Z80 on the old
model 1 computer and i know that this addressing mode doesn’t exist. You would
have to store the buffer pointer in memory and then do the compare to what is
stored there. But i’m not sure as for the 6809.

I couldn’t find it in any of my reference manulas. There that’s about it. Thanks
for the help.

-Mike-

YOU’RE WELCOME .

I HAVE STARTED TO USE YOUR PROGRAM , FOUND SOMETHING STRANGE , I DON’T KNOW IF
YOU INTENDED THIS , BUT , WHEN YOU MIRROR IMAGE THE CHARACTER SET THEN SAVE IT
, THE CHARACTER CURRENTLY BEING EDITED WILL BE MIRROR IMAGED IN RELATION WIR
THE THE REST OF HTE CHARACTER . THAT IS LET SAY I EDIT ‘J’ , MIRROR IMAGE THE
SET THEN SAVE IT . THE ‘WHOLE SET WILL BE MIRROR IMAGED EXCEPT THE ‘J’ .

OTHER THEN THAT , IT’S VERY EASY TO USE .

If timing is not critical, and he’s not too finicky about using
undocumented ROM calls, a simple
LDD ; JSR $BDCC ought to do the trick just fine, printing
the decimal equivalent to the screen at whatever the current
cursor location is.

KO, Marty. That’s really what I wanted to know–whether I could really take
my time and still keep up with the system. I’ve got Leventhal’s book and
I’ll look through it, but not to worry now, right?
Thanks for your help.

Ryan

Sorry about the ambiguity there. I am talking about 0-255, first of all. I
want it to be as fast as possible to give me time to do some other calculations
in between. This is a little control program I am writing for a thermal
exchange system. All it has to do is poll four single-byte locations in
memory, putting their decimal
values on the screen, and put back a single byte value (all between 0 and 255)
into another location which controls a steam valve. Probably I’ll have speed to
spare, but I don’t want to take any chances, and the calculations for the
return value may get a
bit long. I just wondered if there was such a routine floating about
somewhere. ?????

Thanks,

Ryan

Hi, Marty! I just got your binary-to-hex assembly routine the other night,
’cause I am looking for a shortcut through this little problem I’ve got. I
need to know if you know of a wonderfully fast and elegant subroutine to
convert a regular, vanilla binary number (as in Register A, for example) into
the corresponding decimal digit for display onscreen. I can think of all
manner of oblique and slow and awkward solutions, but I just thought that there
might be something out there already (hope, hope, hope)? Know of anything?
I appreciate the info (even if there’s nothing like this available).
Ryan

HI GUYS,

Look, I know you have put great effort into answering questions.
And I appreciate it. GLOEFER: You said to go aheadd and ask
DUMB questions. Apparently that is all I do. You should have
seen me in conference with Marty this morning. If Ya’ll will
stop and think a second you’ll see more of MY problem. For one
example: in the 6821 sheet there is a block diagram on page 6
which refers to peripheral interfaces A & B. On page 7 there is
reference to peripheral data lines. Now I know it sounds stupid
but I missed the significance of the numbers and when I saw
PERIPHERAL INTERFACE in one place and PERIPHERAL DATA LINES in
another, they registered as two separate things. Things like
this drive me crazy. I spent a lot of time going over this
before I caught the numbers and decided that what the author
wrote in two different ways were not two different things. When
an author is so steeped in his own stuff he isn’t always aware
of the fact that such little things can be such huge problems.

MARTY: You thought I sounded dumb in conference this morning.
How about this? According to the RS232 manual a high on bit 7
of $Ff69 is an interrupt. You said the interrupt has to go to
the COCO via the CART line. The CART line seems to be tied to
some bit in $FF23. Assumedly, judging from your positions in
the past, this will be considered little more than lunatic
raving but I can’t make these out to be the same thing. Of
course there has to be an obvious answer. In case you haven’t
figured it out I’m having a little trouble puzzling out what
addresses and bits some of this stuff shows up on when it gets
to the CoCo. I think I’ve come to the conclusion that CRA &
CRB on page 6 of the 682 sheet refer to the entire address set
from $FF20 to $FF23. And that the CART line has to do with
$FF23. It looks like maybe you can write bit 0 to disable or
enable the CART interrupt and that when enabled the interrupt
appears on bit 7. But whether that is true or false I still
don’t know how that connects to the RS232 status register at
$FF69. I think I could write routines to use these addresses
but only for loop scanning, not using interrupts.

Jim

Let me second what I think is the gist of both Greg and Marty’s comments.
That is, for advanced work on any machine, there is and can be no one
“expert” source because even a “simple” machine like the coco has many
different internal components, as well as the very extreme variety of
desires of programmers, hardware hackers and applications. The “best”
source is the find every scrap of technical information on the
particular chips or equipment you are trying the write for. Fortunately,
most of the chips in the coco are from Motorola who provides excellant
documentation, which usually has all the detail you need to use the device,
and usually some good examples or specific simple code to let you
execise it. Clarification or guidance on specific details, which are
unclear to you are about all that can be answered on a SIG
like this one.
(Although this is a VERY friendly and helpful one). The bottom line is
that you will have to do most of the work yourself, but that’s the
FUN part! And don’t feel bad about asking a “dumb” question, we all have.
Sometimes, we know so little, we don’t even realize the scope of the
questions we ask. Whew, anybody left out there in front of the
soapbox? Anyway, start digging and have fun learning! Good luck, Jim.

Gene Loefer

DON,

I’VE BEEN READING YOUR TUTORIAL on interrupts. Guess
I had a misconception about them. I had thought they
were a method of interrupting only when necessary. From what
I read, it looks like just the reverse. Some given routine
is called with each and every instruction. Correct? The interrupts are
continuous and generated by the computer itself.
Not by some input which you can apply to it? I’m looking for
RS232 input during other operations. Is it correct that I
need to check the 232 address via a routine I write to look
at that address – the routine being called at a frequency
determined purely by the rate of interrupt, either 60 hz or
16 KHZ?

Jim LaLone

Thanks for the info, I did finally get the problem fixed. By the way, if you
load EDTASM and change loc $150C to $8C and loc $150D to $1B, it will exit
to the CoCo 3 basic just fine.

Dave Philipsen

You were right! I compared my present version of Super-Patched EDTASM+ to an
old copy that wasn’t patched & found the flaws. About four or five bytes
were screwed up & cause a bunch of problems. Apparently they didn’t exhibit
themselves before.

i have had no problems at all nor with the disk edtasm . i haven7t tried yet
but i think that it should work in the high res text modes too .

The old MPI CAN cause strange effects. I know from experience that
it will cause the coco3 demo to bomb after bouncing ball on my
machine. If you have newer disk controllers that can run without
the mpi, try that. If you have the older style of mpi (-3024 model),
the upgrade only costs about $7, can be done by anyone (mine is
living proof!) I got mine ordered through local “Shack” in about
a week. Other than that, you could try running in the “ROM” mode
by changind all the clr $FFDF (go to ram mode) to clr $FFDE
in the patch assembler code. Good luck. I know I like the
superpatch, zbug combo the best also.

Gene Loefer

I just checked my version . I had no problems with insert, either
new lines or at beginning of old file. i1,1. Have you fixed 64k?
The superpatched vesion will not work correctly with 64k and disk I/O.
The easiest fix is to change all the clr $FFDF s to clr FFDE.
Good luck. I Hope this helps.

Gene Loefer

I believe, Dan . . .

. . . that by “hummingbird” you are using electrical engineer talk for a
quadrature fan module powered electrostatically instead of electomagnetically.

Of course, we can’t expect you to explain it to us in simple terms .

Do you remember that plastic, two-blade fan sold by Dragonfly Software and
made by Piezo Electric Products of Metuchen, NJ?

Well, the blades were moved side to side by a pair of piezoceramic bending
elements and it moved about five cubic feet of air a minute with 0.11 watts
from the AC line. A very interesting device — and, yes, it had a hummingbird
wing quality. I think it is the
one you are recalling.

It was small enough to tuck inside the CoCo and moved air the same way you
would by fanning yourself with your hand — funny, but effective.

Personally, I have never had any need for a “CoCo cooler.”

Of course, as you say a heat sink would probably work just fine for those
folks
who
do seem to have heat problems.

— Jim

Dale, have you figured out what causes the little frazzles to appear on
the left-hand side of the screen when fooling with the GIME registers while
in graphics mode? It is reminiscent of the IBM PC video sync flaw, and the
little frazzles looked awfully familiar. Any way to get rid of them?

– Erik

STEVE,
I’M TRYING TO LEARN ASSEMBLY,
AND I’D LIKE TO BE ABLE TO
SAVE AND LOAD MY ATTEMPTS AND
EXPERIMENTS ON DISK. I HAVE
EDTASM+ ON DISK, BUT IT’S THE
ONE THAT ONLY WORKS WITH TAPE.
I HAVE A COPY OF “SUPERPATCH”,
BUT I DON’T KNOW WHAT IT’S FOR
OR HOW TO USE IT. CAN EDTASM+
BE PATCHED FOR TOTAL DISK OPER-
ATION, OR DO I NEED ANOTHER
ASSEMBLER OR WHATEVER? I’M
TRYING TO GET COMPLETELY AWAY
FROM THAT NASTY LITTLE CASSETE
DRIVE, SO I HAVE MORE ROOM ON
MY DESK. THANKS IN ADVANCE!
-LOREN HOWELL (XENOS)

That 64K ceiling on directly addressable contiguous memory
is THE biggest draw back of the 6809. Pity Motorola
never produces the rumored “Road Runner” chip, which
alledgedly was in part a 6809 with expanded directly
addressible memory. Or a 689XX series. Sigh!

—marty

It’s very easy to use more then 64k on a 6809 if you are using a good MMU and
a true OS like OS-9 level two.

Steve Bjork

I GOT A FREE 6309 AS A SAMPLE AND HANDED IT OVER TO BRUCE WARNER (OF OS9 MOTD
FAME) TO TEST IT OUT. ( I DID NOT HAVE THE TIME, AND HE DID. HE VIRTUALLY
GRABBED IT AWAY!) IT TO BE A DROP-IN RELACEMENT FOR THE 6809. THERE IS
BOTH AN E-CLOCK AND AN ON
-BOARD CLOCK VERSION AVAILABLE. THEY ARE RATED VARIABLY FROM 1 TO 3 MHZ.
PERHAPS THE COCO4 WILL HAVE CO-PROCESSING 68020 AT 12 MHZ AND 6309 AT 3 MHZ –
TO KEEP UP WITH THE TIMES AND STILL HAVE DOWNWARD COMPATIVBILITY..

WOULD BE NICE!!!
– PKW

Hi 6809er,
After a discussion I
about microprocessors, the guys were trying to tell me that the
6809 is a dead chip! That sales of the 6809 will be off the
market within a year. is this true?
It is hard for me to beleive that this would be true,
becuase, afterall Tandy just released the Coco 3.
If the 6809 was to go out of existence, then why did Tandy
waste there money on the Coco 3?
The guys at work gave me an article on CPU’s and indications
were that the 6809 is better than the 6502, because of its
advanced addressing, and multiply instruction
and it also allows a substatial increase in program execution
‘t become available until 1980
, which is later than when the 6502 came out, that the 6809
did not become a @opular cpu then what motorola expected
that it would. The very first Apple Macintosh used the
6809, but Apple saw no one buying it, so they now use a
68000.
I appreciate your response,
Thanks Cocojohn

You made an excellent purchase.

This will allow you to use Mikeyterm at 1200 baud with flawlessly
smooth operation.

Of course, later on you will want to save yer pennies fer a disk drive.
And the needed multipak to go with it!

—marty

Thanks, Milt!

Yes, I’ve heard that SuperPatched EDTASM doesn’t accept arithmetical
expressions as well as does DEDTASM. (Disk EDTASM) Apparently they
fixed that bug before the Disk version came out.

Thanx again!
Don

DON,

THANKS A LOT FOR ALL YOUR HELP WHEN I WAS TRYING TO GET MY EDIT UTILITY DONE.
YOUR LAST NOTE IN YOUR LAST MESSAGE DID IT. I DIDN’T KNOW ENOUGH ABOUT THE
INPUT POINTER SO I SAVED IT AT THE START OF THE ROUTINE AND PUT IT BACK THE SAME
AT THE E
ND. PRESTO, THE RETURN TO BASIC WAS PERFECT. I REALLY APPRECIATE YOUR HELP.
NOW, WHO ON HERE IS THE ONE TO ASK ABOUT SOME DEEP PROBLEMS WITH, CAN YOU
BELIEVE IT, CASSETTE. I HAVE A FRIEND WHO HAS SOME CASSETTE PROBLEMS AND HAS
SUCH TEMPERATURE EXTREMES HE CAN’T USE DISK.

THANKS AGAIN.

TERMITE

I’ve tried twice to send this by mail and no,luck.
Don,

I’m still working on my string editor patch but I’m not sure I can make it. I
am just not proficient enough with assembler. Maybe you can give me a few tips.
The routine is working under a USR call. Of course, an RTS is supposed to
return you to
the next BASIC line.

In the code, a BSR is done (LBSR in my patch code) and then, depending on the
situation, a BEQ (LBEQ) to code lower in the routine than the calling BSR. LEAS
2,S is then done to correct something, I’m not sure just what. Obviously, the
LBEQ doesn’t
balance the LBSR. I’m not too clear on STACK and PC stuff.

A BRA immediately follows the LEAS 2,S. What I’m wondering is, does the LEAS
2,S just make an adjustment and go on into the BRA or does it return you to the
calling BSR or what? It is after this code that my return to BASIC is messed
up. I tried c
learing three bytes in the $2DC buffer, as you suggested, and then tried
clearing the whole buffer – neither helped. I even set the buffer flag at $70.

The code I am referring to is in Extended Basics EDIT. The code runs from $8533
to $86A6 but the BSR I refer to is at $85AB and the returning BEQ is at $85FA.
This BEQ is subject to a carriage return and sends you to $858A, the location of
the LEAS
2,S. Of course, the difficulty is in that my routine replaces part of this,
requiring the LONG codes returning to a different place.

My code replaces EDIT from $8533 to $85AE. Beyond that there is only the one
point of return (described above) that is not a simple RTS. I have replaced the
JMP’s at $859A and $85A8 with the subroutine that copies the EDIT buffer ($2DD)
to the STRI
NG space and a PULS PC. The copy works because I can print the new STRING after
I bomb back into BASIC. But I think there may be something about this LEAS and
the offset back into it from the ROM address that is causing the ERROR from the
BASIC lin
e. If you do not remember, the ERROR comes from the BASIC line that made the
USR call, not the following line. AND, this is only after I have used the copy
routine. Even after using INSERT, CHANGE, etc., I get a good return to BASIC if
I use the “
Q” option for no changes.

TERMITE
Jim LaLone

Don,

I need to temporarily stash a patch in ROM (under RAM mode) to do a long branch
(about $880 neg offset) from the ROM to a routine in my code instead of to
another ROM routine (RAM hooks are out of the question as it is right in the
middle of a subrou
tine). Hand calculating long negative offsets (which no one explains, not even
LEVENTHAL) would be a pain as I would proably have to do it every time I changed
the code. Also, hand calculating would be incompatible with PIC. What at I
really want
to know is:

How can I get Disk EDTASM to calculate it in the regular program assembly -
based on a label in the source? Register X is available and there is plenty of
room for an LBEQ instruction in ROM (6 bytes). I tried doing it in the main
code and STX
ing the code to the ROM address. But I can’t figure out how to set up the
calculation of the LBEQ nnnn (the instruction I want to stash in ROM) from the
ROM address back to the label in my source. I tried using a separate ORG so I
could branch from
the address in ROM but that part of the assembly apparently does not recognize
the label from the other body of code.

I hope I’ve made myself clear and I hope there is a way to aviod hand
calculating it.

HELP!

TERMITE

Well, I guess I could do it, no I don’t want free time, both of them
are so short that it wouldn’t make sense to do it that way. I don’t think
you will find them as impressive as you think, but if you ant….

d

DON,

i’m not sure what you are looking for but some might be interested in one thing
if it can be used. I’ve been playing with the bit banger routines found on
pages 61 and 62 of NOV 64 RAINBOW

TERMITE
{_

Great idea Don !! I’d be VERY interested in such a collection !
Let me know how it progresses !

Thanks !

Art,

I was just studying your assembly language tutorials and would like a little
help on one point. I think I understand what you are saying but, if so, it sure
seems odd. You seem to be saying in A3 that two-dimensional array pointers are
arranged in
memory consecutively according to the first array element.

Please correct me if I am wrong but I take that to mean that if I had 100
records of 5 fields each with the fields in the second element, ( one record =
A$(1,0) thru A$(1,4), etc. ), the pointers would be arranged consecutively in
memory as follows (
assuming DIM(125,5):

First field, records 1 thru 100
Second field, records 1 thru 100
Third field, records 1 thru 100
Fourth field, records 1 thru 100
Fifth field, records 1 thru 100

Then, if a new record is added, it is somewhat out of whack with what has gone
before, although I can’t imagine how it got in the order shown above.

I think my records are stored on disk with each record complete, record 1,
fields 1 thru 5, then record 2, fields 1 thru 5. As I understand it, it would
make more sense to me if the pointers were stored that way. I am trying to
understand exactly w
hat is happening.

A possible explanation just occurred to me. Does the DIM statement arrange the
pointers in memory all in advance with unused pointers in place but empty? This
would seem to make sense to me and would explain why DIM uses up so much memory.
However
, if this is true the recall of a given 5 field record ( A$(3,0) thru A$(3,4)
for instance ) would seem to require skipping through the block for the pieces
identified in the second number in the two dimensional string variable.

How am I doing? Also, is there such detailed information written down anywhere.
There has to be an origin for this stuff somewhere, (and for words like BUMP,
PARSE, POP and so on). I don’t see those in LEVANTHAL (of course I have’t read
it cover t
o cover, I just try to look things up from time to time.)

Sorry if I’m a bother but, as I said, I like to really understand what is
happening and I’m hoping you are just the one to answer me.

TERMITE

Jim LaLone

If you guys are going to program in assembly, program in assembly.
( Don’t just
use BASIC ROM calls for everything, that’s just programming in BASIC. I am
really confused by all the people who say they want to learn to program in
assembly language but don’t want to learn assembly language. As for floating
point, why bother? If you need to crunch numbers, then use BASIC, the speed
difference isn’t going to be that great anyway. Besides, 99.9985% of all the
stuff ever done needs only integers.

By the way, I believe that the CoCo’s BASIC uses some kind of Chebyshev
polynomial expansion. A real shame since everyone knows that CORDIC is
faster, and easier to write. Someday, I’ll upload that SINCOS routine I
did a while back, then you all will see how easy it is. (just as soon as I
comment it about six years from now.)
by the way, would anyone be interested
in a LIFE simulation (cellular automaton) done in 6809 assembly? it uses the
text screen (32×16 resolution) but it updates the screen 16 times per
second. I’m rather proud of it, I did it myself….SCI

(It’s not completely commented, either)

I think using $FF00 would be just as safe an alternative as using $A000 -
it would seem doubtful that Tandy would move $FF00, since it is widely
used both by keyboard checking routines and joystick presses by a rather
large amount of software.

– Erik

On the contrary! Tying yourself to the ROM, while making one
susceptible to possible incompatibilities when ROM versions
change, is actually the MOST efficient use of RAM memory you
could possibly make, since you can use all 64K of RAM,
plus switch in the ROMs when needed, so that you are in effect
using a 96K machine.

To set the record yet straighter, even the now-obsolete
Disk Utility 2.0 was virtually compatible with ADOS. The only
aspect of 2.0 that didn’t work was the L command for loading and
running a program.

It is N
OT uncommon for folks in the CoCo world to replace or upgrade
their ROMs (look at all the 3rd party types who offer such for sale!)
or for folks to buy a second computer for home or office. Thus I’d
say Don HAS a point in this case.

—marty

Greg, I installed DS on my new CoCo with the Disto 80 col card over the weekend.
Seems to be working. But I did see it do a couple of funny things. Want to do
some more checking soon. I have been out of town and out of business for a
month or so. My old grey CoCo died. Now have a new CoCo II. During the next
month or so … before we move to New York … I hope to get a lot of testing
done. How about trying the DM editor. and feeding it into the Shareware
formatter we published recently … or the XWord formatter. I think that would
make a nice system. The DM editor is so easy to use that is a real nice way to
enter text. Add the “dot” commands … and you can format it in any manner you
like. TSEdit would be OK if my fingers didn’t already know the DS diamond. I
just hate to learn to touch type all over again. Yet, a commander at work
really
likes TSEdit … he learned to touch type on it first … and enjoys working
with
the TSWord formatter … except when you include a new file it breaks to a new
page. I hope to look at that problem for him soon and see if I can find a way
around it. Best Regards. Dale

My feelings exactly, Art. While it is now common knowledge that I am a
rather “sloppy” programmer, the fact remains that my junk still manages to
work on almost all hardware/software configurations. Regarding the problems
of disk I/O vectors… Greg wrote some routines that handle OPEN”I”, OPEN”O”,
CLOSE, KILL, and OPEN for EXTEND. These routines have the honor of actuall
working with JDOS. True, this involved re-inventing the wheel, but this was
one of those cases where we felt the effort would be rewarded. (McPaint works
in all ROM combinations, including JDOS and all BASIC ROMS).

Greg also pointed out to me that Radio Shack actually documents the calls for
cassette i/o in the back of the Color BASIC manual as absolute addresses.
These routines are stuck right in the middle of the ROMs ($A7XX or so) and
you are instucted to call them directly there… this will also severely
limit the amount of fooling around that RS can do.

The point was also made that the new CoCo may have either new ROMs or no
ROMs at all. If they were adding new ROMs, and wanted a machine that was
compatible with the old stuff, they would have to keep the new ROMs very
similiar. And if there aren’t going to be any ROMs at all, then it really
doesn’t matter whether or not you do a JSR

($A000) or a JSR $BACA … even
though the first one is one of those “documented” calls, if the ROMs ain’t
there, it ain’t gonna work, eh?

I’m also beginning to wonder if this new machine is going to be more OS9
specific than I’d like. It seems logical that they’d add a “real” OS9
boot routine in ROM… but R/S’s recent craving for OS9 *only* software
makes me wonder whether or not this new machine is going to be compatible
at all with the current one… in which case it would make this “new CoCo”
simply a “new computer” … a nice OS9 level II machine (though “nice” and
“OS9″ really don’t go together in my book).

.. “And amongst the fields shall find the golden dice, wherewith we played
of yore.”

– Erik

Art, You are in for a BIG “GOT-YOU”!

The two major problems with using ROM calls for the random number generator
are the lack of compatability and the slowness. For example, I did a life
in assembly (low-resolution, just 32X16) I needed a random number generator
to start the “universe” and I did a quick and dirty (mul 13, addd timer)
RNG that was adequate for my purposes, the overhead of converting from
integer to floating, doing the generation, then converting back to integer
seems to me to be rather inefficient, but then I did a sin and cosine routine
in integer math just to see if it could be done.

Yep! did de job so well, that IF the rest of the experiment works,
I will modify the nice graphics you did to support
a normal disk speed of 180 rpm. Unless YOU want to help me!

heh heh!

—marty

Well, Steve,,
I did a quick and dirty fix of your Disktimer program to make it
NOT crash in the graphics department when it was used to measure
a disk rotating at 180 RPM. I got the drive ajusted to 180 RPM
very smoothly using your program. No problem at all.
Later, if the rest of the experiment checks out
(if teh signal strength and information format… sector headers
and the like) are compatible with the WD 17XX format on the
IBM PC AT disks) I’ll do a more thorough job of altering
Disktimer to make its grapics displays compatible too.
Rite now I just use the number, and disabled the graphics
and the out of bounds checks in statement 730 and 760
More later! I will tell you if we ARE able to read PC AT disks
on a CoCo using NON PC AT type 80 track drives set to 180 RPM.
The speed seemed VERY stable when I made the needed minor mod…
so I am hoping for the best. But as I and you know,
problems with signal strength, head resolution, and information
format on the disk could still cause this little experiment to be
a failure. We shall see. At the moment I am using a Tandon TM 100-4
80 track double sided drive. Full hieght. But have full
tech references for Teac, Phillips, and Shugard half height
80 trackers… and the needed alteration there to make THEM
180 RPM capable are pretty easy, it appears.

—marty

Here is a subroutine that uses no rom calles (great for 64k programs or OS-9).
The random number from 0 to 255 is returned in the A register, no other
registers are changed.
And here it is…..

RANDOM PSHS B
LDD RND
LSRA
EORA RND
COMA
LSLA
LSLA
LSLA
LDA RND
ROLB
ROLA
ANDA #127
STD RND
TFR B,A
ADDA SEED
EORA RND
STA SEED
PULS PC,B
RND FDB 0
SEED FCB 0

Well, there you have it. I will be uploading it into the Asm Lag Database with
full documentation in the next few days.
Use any number (besides $FFFF) to reseed the sequence (it’s length is about
32,000 numbers long).
Steve Bjork, 6809er

To generate a random number, use the following procedure:

LDD #MAX the maximum value you want to have
JSR $B4F4 copy D into FPAC 1
JSR $BF1F generate a random number
JSR $B3ED retrieve FPAC 1; D= your random number

For example, to generate a random number between 1 and 10:

LDD #10
JSR $B4F4
JSR $BF1F
JSR $B3ED

And D will now equal the random number. You may want to do the equivalent of
a R=RND(TIMER) instruction, thus “randomizing” the RND seed… Do a:

LDD $112 timer value
JSR $B4F4 put TIMER into FPAC 1 for max value
JSR $BF1F generate a random number

Is it live…. or is it Milliluk?

– Erik

Sorry to hear you will be gone for a while… tho note that
this info service thing DOES cost, and it is necessary to
pace one’s self!

—marty

What I propose instead is that we get to you an alternative terminal
program that wil do all you need automatically. Greg Miller here seems to have
writtensuch,
and we hope to hae it up here soon.
But i wil try to get you a few representative sets of addresses.

—marty

BY THE WAY . . . IF YOU HAPPEN TO CALL IN BEFORE YOU ARE READY TO DOWNLOAD-
PLEASE LEAVE ME YOUR CIS ID # SO I CAN RUN A MORE THOROUGH SEARCH FOR TELETERM
(TAPE VER.) ON THE COMPUSRVE SIG. THANKS…

—LANCE(STARCOWBOY)

GREG- GREAT TO HEAR TELETERM MIGHT HELP ME OUT HERE! HOPE TO SEE IT IN DAT
SOON.
BY THE WAY- IS IT ALREADY IN THE COMPUSERVE SIG? IF SO- WHAT NAME?
THANKYOU VERY MUCH.

STARCOWBOY (LANCE)

How curious! As I said, I never really got very educated
about cassette tape format.
I did of curse know that CSAVEM did not support segmented files
(but then again SAVEM does nto suport segmented files either!)
but did NOT know that Ext Basic would recongnise and read a segmented file. How
odd!

Mind Reader! (assume you just read the message I just left you
before this one!!)

How to suport both?

Just use disk binary format as standard…

check file when it is in buffer for presence or absence of segments…
if no segments (other than one) then convert to tape at level of the
terminal program.

before saving.

Tape format for the CoCo ONLY suports non segmented files.
Worse yet, as Art Flexser points out,
the byte for byte Disk image
is different form the byte for byte tape image, for the latter is in the form of
256 byte long
blocks with check sums.

But, IF one restricts oneself to non segmented ML files,
one should be able to
(albeit with some hassel in writing the program) support
properly buffer saves of
ML programs to EITHER disk or tape.

Do ask Greg to take a look at this thread, please. Thanks!

—marty

Good Point… It has been so long
since I messed with tape I’d forgotten that its format
is different even from
non segmented disk format!!

Well… it still would
be possible to support
proper saves to tape
of non segmented files.
Tho forsure, a hassel!

Could you ask Gweed
for permission for us
to have that TAPCNV util
here? You see him more often
than I do… and it seems
like it is just what this feller needs!!!

Thanks!!

—marty

It’s not that its all that hard to add to the description those addresses,
the difficulty falls in the fact that 80% of the people who upload these
files won’t give the addresses required or will give incorrect addresses, and
in the problem of the MASSIVE task of changing all the descriptions in the
existing files of the database.

I think the Teleterm solution may very well be the best one. Tho, of course,
we’ll have to wait to see whether an independent taste test shows that
Teleterm
works sufficiently with Delphi to be used.

– Erik

Art,

TAPCNV is the 11th program in the set of files for MIKEYTERM here on Delphi.

Cray

Oh Ho!!! Sorry… have NEVER used MTERM with cassette ever!

I will check it out later (I do not have a cassette recorder
here where I am now… only a disk drive equipped CoCo-ized Dragon and
an 80 col card).
But I still find it very odd that MIke would NOT allow you
to automatically save cassette
fies in binary ML format…
unless he was concerned about
trying to save segmented disk
files to cassette (which is impossible as segmented binary
files do not EXIST on tape, tho they do on disk).

Sorry about the confusion.

Hmm… I just checked out the
prompts, and YES.. the durn program does ask for start
and execute addresses!!

HO
W very odd!!!

He really could have done better than that.

He could have automatically checked for single block
vs segmented binary ml format,
allowed automatic assigning of
proper start, end, and exec adrs
in the former case and a warning message in the latter case.

Veryt Very odd indeed!!!!

With much regret I must say that considering you are the
FIRST person here to bring up this issue as a problem,
and considering the many HUNDREDS of man hours it wouold
take to add the info you need
to our file descriptions, I must
decide that we will NOT be making
the changes you request.

Perhaps other (commercial) programs will
more properly support X modem DL’s of binary files for the CoCo…

Short of that I am afraid you
EITHER must develop utils for
converting such files into
proper tape files (all the data you need… start and exec adr..
IS in the file you down load…
a quick study of the Disk Basic
Unravelled manual…or of my
article in the February Rainbow
about disk file formats…
will tell you WHERE that info is).

sorry!

marty

Both your messages make me wonder if we are talking abut the SAME PROGRAM!

I’m talking about MIKEYTERM (Ver 4.0, tho what I have
to say about it is true of ver 2.3 and up).

You save Machine Language program file AS Machine Language program type.
There
is never a prompt for Hex Addresses
at any point in the upload or download process.

What may in part be confuding you is that a ML Program
(the kind of file that has start, end and execute address info as
a critical part of it) is only one of
three binary file types supported by Mikeyterm.
Mikeyterm also supports Tokenized Basic (a very common binary file tupe used
whenever you save a BASIC program to disk) and a file type called “binary data”
which is provided for by
Radio Shack Basic but nearly never used by existing Coco softtware.

—marty

Please read your documentation. You are completely mistaken.
As a DAILY user of Mikeyterm for the last several YEARS, I can
assure you that hex start, end, and execute addresses are in
NO WAY needed for Mikeyterm up or down loads. I’ve been using
that program from the days of its beta test early ancestor,
Mikeyterm 1.6. I have no idea where you got that completely
WRONG idea!!

—marty

Yes, provided that the 40-track disk is not completely full. OS-9 will
read the first sector and find out that the disk you are reading is 40-track.
To copy the files from it to the 35-track disk, you need to use dsave to copy
a directory at a time…or the entire disk. Good luck.

Art,
I want to thank you for answering my question so quickly, but if I had known
that you were on Delphi I would have addressed the question to you. I have been
using ADOS for the last year and a half and think that its the greatest. Could
you tell me wh
ere to put a JSR in ADOS ot get it to auto execute the DOS command on powerup? I
live in an area that the power either flikers or goesout at least twice a day
and I use my Coco remotely from both school and work(I am currently a student
attending UF
in Gainesville). Thanks in advance.
Bob Thomas

I hate to tell you this but the color computer can only speed up when it’s
reading ROM. The Ram is already double timing with the CPU and VDG!

Steve Bjork, SRB Software, 6809er

Didn’t get all your message. At least didn’t see a name. Over the course of
three or four months last summer and fall we ran short patches that appraoched
the single drive C compiler problem from several different angles. They all
work.

But, it sounds like you have already found that out. By the way, you can use
a ramdisk to hold the temp files and your operation will speed up many fold.

Or, you could use the ramdisk to hold the compiler system disk. This would
save you a lot of time also. If you can’t find the specific articles with
the patches let me know and we’ll try to look them up.
dale

Ron, you stumped me on this one. Not too much of a hardware type here.
Suggest you pass the same question email to Brian Lantz. He may be able to
figure it out.

Sorry, we couldn’t help.

dale

(1) system emulators are very sophisticated… and VERY powerful tools.
A long time ago I considered bringing up a 6809 emulator on the CoCo
to (of all dumb , trivial things)help me break protected ROM packs.
But never did it. A friend wrote a Z 80 emulator on the CoCo
and nearly borught up CPM via emulation on his CoCo!

(2) BTW… have you seen the Steve Bjork program Coterm???
Do you think YOU could make a “version of” coterm that used
the BIT BANGER port? What I refer to is a “split screen”
dedicated Converencing utility, that gives you one (in the case of
Steve’s program) or two or three or four (in the case of this program
Q modem I am using on my PC clone) lines of protected space on the screen
into which you can type, and what you type there is NOT chopped up
by any incoming messages (which all appear in the upper part of the screen.
)
You can edit your line in the protected area (primarily with back space /
delete… tht realy is all you NEED for a single line worth of edting)
and it is NOT transmited out the 8{ort until you hit C/R.

I bring this up because there are lots of half decent CoCo terminal
programs… but NOT split screen utils for the CoCo for converencing
for those folks who have neither a RS232 pak nor a PBJ word pak.
(Steve’s COTERM requires an RS232 pak at least, and a version of
it supports the PBJ pak). So, while I suspect ifyou make
a full featured smart terminal program you will make a VERY
good one, perhaps as a quick side project you could make us
a varient of that EXCELLENT program you sent me that suports
this split screen option. For the many folks who want such
capability but do NOT have the wierd extra hardware required
by Steve B.’s COTERM.

Steve B. MIGHT even consider giving you his surce for COTERM…
tho I doubtyou’d need it, as you seem to have the
business of terminal software well in hand.
Note that this project would involve some care in buffering
keyboard input while one is editing one/s message on
the bottom part of the screen so that one does NOT lose any
incoming data during such actions.

Use of interrupts might be handy.

I must say that the little bit of user interfacing you did
in your terminal program (for setting prarmeters) was, in my
VERY critical opinion. VERY well layed out. You reely
seem to be QUITE good at this sort of thing. I was truly impressed.

David,

The source code for Mterm is not available for distribution.

As for 1200 baud, I highly recommend the RS232 pak or the PBJ-2 SP.

Mike

Why not just change the address (and your serial chip) to match the
Radio Shack Deluxe RS-232 pak. In this way you will not ever need to change
any software you get! Radio Shack uses a 6552 at ports $FF68 to $FF6B.
Steve Bjork, SRB Software (6809er)

The Brian Lantz utility ‘CAT’ published in the Dec
85 issue of rainbow doesn’t appear to work as it
appeared in the magazine on page 262-263.

I had to add two lines of code to get everything
to work as Brian suggests. The 2 lines are listed
below with a large portion of the original source
for your ease in locating the new lines. They are
marked with ‘**** in the comment field ****’.

I haven’t exhaustively tested the routine but all
appears to function well. I’ll be adding line
numbering and a little more error guidance soon.

bsr optchk process options
stx prmptr store X at PRMPTR
clr ipath set default path (0)
lda 0,x get a character
cmpa #$0d is it ?
beq main if so, skip to main
*
lda #read. set access mode for reading
os9 i$open attempt to open file
bcs exit if an error, end program
sta ipath store input path #
stx prmptr store parameter position
*
bsr optchk ****REQD to set RANGE****
stx prmptr ****REQD to END OKAY ****
*
* This is the main loop
*
main lda ipath get input path #
leax buffer,u point to buffer
ldy #bufsiz set ‘y’ to max # of bytes
os9 i$readln attempt to read a line
bcs eofchk branch out if error or EOF
ldd count get current line count
addd #1 add one to the line count
std count store the new line count
cmpd from compare COUNT to FROM

See ya….
Vince Caldeira
(VINCECALDEIR)

Jeff is hardly the only one capable of writing good user interfaces..
indeed, we both kno who gaveJeff the idea for both Disk Util
and for OS9 Solutoin, don’t we, eh? I suspect that the folks
at Fort Worth are just too lacking in intelligence and vision
to realize how important it is to have such stuff.

Two friends of mine literally could NEVER have used teh CoCo if it
was not for Jeff’s Disk Util that I had burned into their ROM’s.

Jeff’s (YOUR) OS 9 solution is a good program… but actually
what is even more needed is a module like that than can be called by
all OS9 application programs, and implimented either in full or in
part whenever such programs need to do Disk I / O.

That way All OS 9 application programs would have similar high
quality user interfaces, and software developers would be saved
massive amounts of time and headaches.

I HAVE source for SIGMON in Macro 80C format…
indeed, I’ve had source for SIGMON for years.
I aliso have source for “DISKMON”, a moderatly kludgy but
still useable modification of SIGMON that allows
SAVEM and LOADM (including offset load suport)
direct from SIGMON.
Indeed, we “created” source for SIGMON (by disassembling the
program to MAG media and then laboriously typing in the
hard copy of the source code that Data Soft actually published
when it sold the program.

The main argument AGINST using SIGMON is that its legal status is
very much in limbo, and it might be VERY hard to find
and persualde whoever would have the authority allow you to
co commercial with an OS9 SIGMON. Which is a GREAT pity, because
SIGMON is SUCHJ an outstanding product.

Tandy has been disappointed with how poor the third party
support of OS9 has been, but with such utter garbage
for “tools” as the stuff they include with it, it’s no wonder!
Now, if they had provided modular user firendly
(ponit and pick, a al Jeff Francis’ RS DOS Disk Utility)
Disk I/O routines, and modules for sleecting stuff from
a menu of choices, and the like, that would have
sparked the development of a variety of
quality software from a lot of different corners…
whihc all would be similar in control structure
and so easy for users to learn. But they lack the
intelligence and vision to do that… and their
lack of intelligence and vision… the same lack that
releaged the CoCo to a few dedicated hobbiests…
will most likely kill OS9 too. Pity… I was just
getting to appreciate how POWERFUL that OS is…
despite it repulsively trashy interface.
I even bought a copy of it recently on the half price sale.
Figgerred the four manuals were nearly worth the $35.
(yeahg.., I know the upgrade to ver 2.0 is on its
way tomy store.)

YES, YOU SHOULD HAVE USED XMODEM TO DOWNLOAD. IT’S THE MOST ERROR FREE WAY TO
DOWNLOAD.
THE ’128′ IS JUST THE EXTENSION AND LIKELY REFERS TO A TAPE COUNTER POSITION OR
A PAGE NUMBER IN A MAGAZINE
THE BINARY MEANS IT WAS MACHINE LANGUAGE AND WOULD HAVE TO SAVED AS SUCH WITH AN
EXTENSION OF 128, HOWEVER IT COULD BE RENAMED LATER TO SUIT YOUR NEEDS.

I will be writing and posting a review of the opular terminal programs.
I’ll be most interested in your comments on it when it is done.
At present Mikeyterm seems likely to come out as #1. Tho I have
yet to see the current issue of Color Connection.
(It’s on the way to me now).

I HAVE DOWNLOADED MTRM40 AND CONFG40. AFTER CONFIGURING MTRM40 ,I TRIED TO USE
IT BUT IT HANSUP AFTER DIALING NUMBER. ANY SUGGESTIONS. I HAVE 232PAK. THANKS
JACK

Would you be intrested in an updated
version of you basic dissasembler?
I have modifide it so it dissasembles
straight from disk. I want to add
a few features like defining tables
and send it to an output file, but
god only know how long that would take.

Does anyone remember which Rainbow had the
Rom calls for disk I/O?

I’ve written a neat terminal program for the cococo rs232 pak 300/1200 baud that
gives
you a beatiful 40 column screen but I
would like to add xmodem. If someone has
a good description of the protocal or even better
a assembly listing please let me know. I will try to upload this terminal
program
when I’m done.
thanks

I TOO WOULD LIKE TO SEE THE SOURCE AVAILABLE! BEING FROM CANADA UNFORTUNATELY
WILL BE CAUSING SOME DIFFICULTIES. I SEM
EM TO GET BUGS WHILE ON-LINE. MOST DISTURBING. I AM USING MIKEYTERM. WILL HAVE
TO TRY SOMETHING ELSE LATER. ANYHOW BACK TO DATAPAC. I REALLY HOPE THAT SOURCE
CODE IS PUT ON-LINE FOR DOWNLOADING. I TRY TO READ OVER ANY SOURCE IN THE
RAINBOW (EVEN
FOR SOMETHING OF NO USE TO ME WHATSOEVER) FOR THE AUTHORS TRICKS, SHORTCUTS ETC
ETC ETC. WEL

WELL, THANKS FOR CONSIDERING THE IDEA. AND PLEASE EXCUSE THE TYPOS. I JUST CAN’T
SHAKE THEM FROM THIS SYSTEM.

PAUL STILWELL (NEPAS)