Errors 48 ROM
ZXNet echo conference «real.speccy»
Interesting information from the book "Caches of the zx spectrum":
===============
9. ERRORS IN THE SYSTEM
THE AUTHORS OF THE OPERATING SYSTEM AND THE BASIC INTERPRETER
ZX SPECTRUM DID EXCELLENT WORK, BUT WE CAN'T
AVOID SEVERAL ERRORS
9.1. DIVISION ERROR
UNDER ADDRESS #3200 THE VALUE #E1 IS PLACED INSTEAD OF #DA. AS A RESULT, THE LAST BITT IS SOMETIMES LOST, WHICH RESULTS IN ERROROUS ROUNDINGS. THE CONSEQUENCES OF THE ERROR IS DEMONSTRATED
PROGRAM:
10 LET A=A/B
20 IF A THEN GO TO 10
30 PRINT "RECEIVED NULL"
RUNNING IT WITH THE VALUES A=1 AND B=3, AFTER A SECOND WE WILL RECEIVE
PRINT: "RECEIVED NULL". RUNING IT AGAIN WITH THE VALUES
A=1 AND B=2, WE LOOP, BECAUSE SPECTRUM BELIEVES THAT
2^-128=2^-128/2.
9.2. ERROR "-65536"
THE AUTHORS MADE AN INACCURACY IN THE REPRESENTATION OF THIS NUMBER. ONE TIME IT IS STORED AS A FLOATING POINT REPRESENTATION, THE OTHER TIME IT IS STORED AS AN INTEGER IN 2'S COMPLETION CODE. THE CONSEQUENCES OF THE ambiguity APPEAR FOR EXAMPLE WHEN EXECUTION DIRECTIVES: PRINT - 65536. THE NUMBER -1 WILL APPEAR ON THE SCREEN.
9.3. ERROR CHR$ 8
THIS CONTROL CHARACTER SHOULD MOVE THE CURSOR ONE POSITION TO THE LEFT, OR TO THE ENDPREVIOUS LINE. THIS IS EXACTLY THIS IS HAPPENING IN LINES 1...23, BUT FROM THE BEGINNING OF THE 1ST LINE
TO THE END OF THE 0TH LINE IT IS IMPOSSIBLE TO GO TO. FOR INTERESTING
EFFECTS, PLEASE MOVE THE CURSOR TO THE LEFT OF THE FIELD
(0,0).
9.4. ERROR CHR$ 9
THIS SYMBOL SHOULD MOVE THE CURSOR TO THE RIGHT ONE
POSITION. HERE, HOWEVER, A SERIOUS ERROR IS MADE: ALL THE NECESSARY CALCULATIONS ARE PERFORMED, BUT THE AUTHORS FORGOT TO MODIFY THE SYSTEM VARIABLES.
9.5. ERROR "PRESS ANY KEY ..."
IN SOME SITUATIONS SPECTRUM INTERRUPTS OPERATION AND WAITS FOR
THE USER PRESSING ANY KEY. THE ERROR LEADS TO
THE COMPUTER DOES NOT RESPOND TO THE CS OR SS KEY,
ALTHOUGH IT RESPONSES TO BOTH AT ONCE.
9.6. TICKET LINE INDEX ERROR
Suppose that the last line in the program is numbered
1000. WHEN YOU PRESS 1001 AND "FNTER" AND THEN CS/1, LINE 1000 WILL BE SCAN TO THE BOTTOM OF THE SCREEN, BUT TOGETHER WITH THE TICKET INDEX WHICH IS BEFORE WRITING LINES
MUST BE DELETED.
9.7. DELETE ERROR
WHEN DELETTING CONTENTS IN THE BOTTOM OF THE SCREEN USING
CS/1, THE CURRENT LINE OF
THE PROGRAM IS SCAN INTO THE BOTTOM OF THE SCREEN AND IT IS NOT POSSIBLE TO RETURN TO THE CONTROL SIZE THIS
AREA. YOU NEED TO PRESS "ENTER" AGAIN.
9.8. LEADING SPACE ERROR
SOME KEYWORDS INDISPLAY TIMES ARE NOT
ALWAYS SEPARATED BY SPACE FROM PRECEDING ones: TRY
FOR EXAMPLE, RUNNING
PRINT CHR$ 255;CHR$ 13;CHR$ 255
9.9. ERROR K MODE AFTER PRESSING THE KEY IN K MODE AND HOLDING IT IN
THIS POSITION, THE KEY STARTS TO EXPAND. THE CURSOR
CHANGES TO L OR C, BUT THE SYMBOL IS PRINTED IN K MODE ALL THE TIME.
9.10. SCREEN$ ERROR
CELL #2570 SHOULD BE #C9 INSTEAD OF #C3. AS A RESULT, THE PROGRAM WILL OUTPUT 22 INSTEAD OF THE EXPECTED NUMBER 12:
10 PRINT "1234567890"
20 LET A$=SCREN$(0,0)+SCREEN$(0,1)
30 PRINT A$
EVEN MORE AMAZING WE WILL GET A PRINT IF AT THE END OF LINE
20 WE ADD +SCREEN$(0.2)+SCREEN$(0.4). VARIABLE SIGN
VALUE "55". THIS IS AN EASY ERROR. ADDING TO A$ THE VALUE OF SCREEN$(0,1) IN ALERTS, NOT IN ONE EXPRESSION.
9.11. ERROR STR$
WORKING WITH NUMBERS -1 9.12. ERROR CLOSE AN ATTEMPT TO DISCONNECT THE FLOW 4..15 FROM THE CHANNEL BEFORE ITS CONNECTION LEAD TO UNINTENDED EFFECTS WITHSYSTEM RESTART 9.13. THE RET ERROR IS ONE OF THE HARDEST ERRORS TO LOCALIZE. IT APPEARS 9.14. NMI ERROR THIS IS ONE OF THE SERIOUS ERRORS. SWITCHING ONE BIT AT ADDRESS #006D CAUSED THE FACT THAT SPECTRUM IS NOT ABLE TO ACCEPT NON-MASKED INTERRUPTIONS FROM THE USER PROCEDURE BY SERVICE PROGRAMS. SUCH INTERRUPTS CAN EITHER IGNORE OR RESTART THE SYSTEM BY GOING 9.15. PAUSE N ERROR THE PROBLEM WITH THE PAUSE INSTRUCTION IS THAT 9.16. CLS ERROR WITH SYSTEM VARIABLE DF_SZ < 2, IT IS DIFFICULT ============================= * Crossposted in CODE.ZX [I ZX] With warm greetings, Slava! [We don’t need the sun - the party is shining for us, we don’t need bread - work come on!]
PRINT "ALA"+"BUM CUK CUK"+STR$ .0.01
PRINT 7+VAL STR$. 0.01
IN BOTH CASES ONLY .001 WILL BE ON THE SCREEN, THE BLAME LIES WITH
SPECTRUM.
INCLUSIVE. AND ALL BECAUSE IN THE ROM, IN THE TABLE CONTAINING
DATA ABOUT CHANNELS WITH ADDRESS #1716, THEY FORGOT TO PLACE THE INDEX-
END OF THE TABLE.
SOMETIMES WHEN RETURNING TO ZX-BASIC FROM A PROGRAM IN MACHINE CODE, CALLED THROUGH USR K. FORGOTTEN BEFORE RETURNING TO THE INTERPRETER, RESTORING A PAIR OF REGISTERS H'L'. IF THEY WERE USED BY YOUR PROGRAM AND MODIFIED, THEN WHAT WILL HAPPEN WHEN YOU RETURN TO ZX-BASIC IS A MATTER OF CASE. USUALLY
THE SYSTEM HOLDS, BUT NOT ALWAYS.
TO ADDRESS 0. THIS MAKES A CONTROLLED RESTART OF THE SYSTEM IMPOSSIBLE AFTER ITS FAILURE, AS WELL AS THE IMPLEMENTATION OF OTHERS COMPUTER CAPABILITIES. ELIMINATING THIS ERROR REQUIRES DIFFICULT
AND COSTLY IMPROVEMENTS.
IT IS NOT ALWAYS EXECUTED. AN ERROR IN THE PROGRAM THAT SERVES DUPLICATION OF KEYS RESULTS IN THE FACT THAT WHEN YOU PRESS ANY KEY IMMEDIATELY BEFOREBY executing the PAUSE STRUCTURE, IT IS IGNORED. EXAMPLE:
10 PRINT "RELEASE THE KEY WHEN YOU HEAR A SOUND"
20 FOR 1=1 TO 500; NEXT I
30 BEEP 1.10
40 PAUSE 0
50 PRINT "END"
THE PROGRAM BEHAVIOR DOES NOT EVEN HINT TO THE EXISTENCE OF LINE 40 IN IT.
TO TAKE THE RESULT OF THIS COMMAND AS CLEARING THE SCREEN. FOR
DF_SZ = 1 THIS CAN BE CONSIDERED IMPORTANT, BUT FOR DF_SZ =
0 THIS IS ALREADY A TRAGEDY.
WE HOPE THAT THE LIST ABOVE EXHAUSTS THE “NEGOTS” WHEN PROGRAMMING ON THE ZX SPECTRUM.
* Crossposted in REAL.SPECCY
Hello, All!
Is there a version of the 48th ROM with corrected errors, and how does
existing software behave with it?
All the best.
Best regards, Roman.
Here one day, September 3, 1999, Vyacheslav Mednonogov and All were whispering, but I still heard everything! They said, they say,
VM> Interesting information from the book "Caches of the zx spectrum":
VM> ===============
VM> 9. ERRORS IN THE SYSTEM
A review of ROM errors was also in the "ZX-Review" for 1992, but more
complete and more sensible.
Good luck!
Vitalic.
... *That's it!* Inscription on the cemetery.
VS> A review of ROM errors was also in the "ZX-Review" for 1992, but more
VS> is complete and more sensible.
So throw them in and we'll see. ;)
Long life and prosperity to you, Vitalic!
With best wishes, Musa Khaybullin.
RM> Is there a version of the 48th ROM with corrected errors, and
RM> how does the existing software behave
RM> with it?
there are several versions, for example 1990 and 1991. The best, in my
opinion, is the 1991 firmware, because... For example, it has already fixed the work with
non-masked interrupts. and there are also several new bells and whistles
With best wishes, Max Shashkin.
VM>> =============================
VM>> 9. ERRORS IN THE SYSTEM
VS> A review of ROM errors was also in the “ZX-Review” for 1992, but more
VS> complete and more sensible.
MK> So throw them in and we'll see. ;)
It’s too hard to stuff. True, if the crowd gets too crowded,
I might get tense... :)
Good luck!
Vitalic.
RM>> how does existing software behave with it?
MS> there are several versions, for example 1990 and 1991. The best, in
my
MS> opinion, is the 1991 firmware, because... For example, it has already fixed
MS> work with non-masked interrupts. and there are also several new
MS> bells and whistles
Once upon a time, I even tried to install it myself.
But then there was a tendency to install the native, 1982 ROM, they say
programmers have adjusted to the errors and the corrected ROM will generate glitches in the
software.
So I’m asking, maybe someone has fresh experience with the latest software with
a similar ROM?
And my NMI error was fixed a long time ago.
MS>> work with non-maskable interrupts has been fixed. and there are also
MS>> several new bells and whistles
RM> Once upon a time, I even tried to install it myself.
RM> But then there was a tendency to install the native, 1982 ROM, they say
RM> programmers have adjusted to the errors and the corrected ROM will generate
RM> glitches in the software. So I’m asking, does anyone have any recent experience with
RM> new software with a similar ROM? And my NMI error was fixed a long time ago.
I don’t know the latest software, and some moronic bootloaders didn’t want to be friends.
I’ve seen about 5 such programs. Only this was a long time ago, in those days
when they still sent fucks to them.
And NMI is more fun there. It went into the monitor that was located in the free space. At least I really liked it...
Sincerely,
Renat
RM> they say programmers have adjusted to errors and the corrected ROM will
RM> generate glitches in the software. So I’m asking, does anyone have any recent
RM> experience of working with new software with a similar ROM? And my NMI error
RM> was corrected a long time ago.
And the calculator error "-65535-1" was corrected?
It was not there in the ZX-81, but in the ZX-SPECTRUM it appeared from somewhere :-/
Kirill Frolov. [ZX]
KF> Did you correct the calculator error “-65535-1”?
No.
I’m thinking, maybe we should dig into it already?
Which version of the error is corrected, but doesn’t have any bells and whistles?
In the ROM of 91 and the reset, it seems, remade?
RM> I’m just thinking, maybe we should dig around there already?
Enough, you either need to write a new code or not change the old one, and working with ROM
at this stage of development is _UNACTUAL_
RM> Which version has the errors fixed, but doesn’t add any
RM> bells and whistles?
in any
RM> In ROM 91 and the reset seems to have been redone?
redone, but does that change anything??? anyway, after all, you first go to BASIC 128, and all the modifications are to BASIC 48.
MS> has been redone, but does that change anything??? all the same, after all, first
MS> you come out
MS> in BASIC 128, and all the alterations are in BASIC 48.
I don’t have BASIC-128 for a long time, it’s NeOS. And pressing the reset button enters MADROM.
And I also want to put things in order in the 48th part of the ROM. :)
Such firmware is included with Yudinov’s zxemul... Just install the normal font
.
Kirill Frolov. [ZX]