Mad Pascal

tebe · February 19

fixed, https://github.com/tebe6502/Mad-Pascal

roeoender · February 20

Great, thanks for quick fix!

This unearthed another thing that seems to be semicolon handling bug, that took me long to replicate and identify the cause.

In attached code crucial part is in wpn.pas - legal no semicolon after CRT_Readkey prevents working of "targetHpPtr^ := 0;" (but only when assignment is to 0).

	if (targetHpPtr^ > g_baseIdx) then begin
		targetHpPtr^ := targetHpPtr^ - g_baseIdx;
		CRT_Write(' RESULT:'~);
		CRT_Write(targetHpPtr^);
		CRT_NewLine;
		CRT_Readkey // no semicilon here is legal but causes generation of incorrect code! only affects assignment of 0, other values would work correctly
		// adding semicolon generates correct code
	end else begin
		targetHpPtr^ := 0; // doesn't assign!
		CRT_Write(' EXPECTED 0:'~);
		CRT_Write(targetHpPtr^);
		CRT_NewLine;
	end;

Change can be seen in generated assembly code (in 2 places)

Bad code:

; optimize FAIL ('B_CRT.CRT_READKEY', WPN), line = 21

	jsr B_CRT.CRT_READKEY
	inx
	mva B_CRT.CRT_READKEY.RESULT :STACKORIGIN,x
	dex
	jmp l_0257
l_0239

; optimize OK (WPN), line = 24

	mwy TARGETHPPTR :bp2
	ldy #$00
	sta (:bp2),y

Wrong output:

INI HP:5 DMG:1 RESULT:4
INI HP:4 DMG:1 RESULT:3
INI HP:3 DMG:1 RESULT:2
INI HP:2 DMG:1 RESULT:1
INI HP:1 DMG:1 EXPECTED 0:1
INI HP:1 DMG:1 EXPECTED 0:1

After adding semicolon I got correct result:

INI HP:5 DMG:1 RESULT:4
INI HP:4 DMG:1 RESULT:3
INI HP:3 DMG:1 RESULT:2
INI HP:2 DMG:1 RESULT:1
INI HP:1 DMG:1 EXPECTED 0:0
INI HP:0 DMG:1 EXPECTED 0:0

And correct asm:

; optimize FAIL ('B_CRT.CRT_READKEY', WPN), line = 21

	jsr B_CRT.CRT_READKEY
	jmp l_0257
l_0239

; optimize OK (WPN), line = 24

	mwy TARGETHPPTR :bp2
	ldy #$00
	tya
	sta (:bp2),y

pascal_semicolon.zip

tebe · February 20

3 hours ago, roeoender said:

This unearthed another thing that seems to be semicolon handling bug, that took me long to replicate and identify the cause.

fixed

roeoender · February 21

8 hours ago, tebe said:

fixed

Thank you! Now works correctly also in my main project.

Now on my main project I stumbled on a "gotcha" that could be prevented with a warning, in attached program file dd.pas defines:

const COLOR_GREEN = 2;
const GREY = 1;

Which run.pas tries to simply print:

	writeln('COLOR_GREEN ', COLOR_GREEN);
	writeln('GREY ', GREY);

And the result is quite unexpected:

COLOR_GREEN 196
GREY 8

The base reason is that system.pas and ctr.pas define such generic-named constants and they take precedence (MP documentation warns about order of uses so it is somewhat OK).

This is easy to identify in such simple program, but in my main project MP didn't give me any warning and amazingly program worked differently on an earlied MP version - few weeks before -

it seems that in mad pascal updates since around 2024-02-01 something changed in resolving priority of such redefinitions,

this change may be perfectly OK and justified, but it would be a lot better if redefinition to a constant would be logged as a warning.

Also would be better if constants with simple names had some name prefix to prevent such clashes - seems that Bocianu's blibs to some extent follow such convention examples:

'PMG_gractl_missiles' in b_pmg.pas, 'CRT_screenWidth' in b_crt.pas

I understand that there may be no easy/backwards compatible solution to this.

pascal_redef.zip

tebe · February 21

25 minutes ago, roeoender said:

Now on my main project I stumbled on a "gotcha" that could be prevented with a warning,

USES has recently been improved so that it works as in FPC

uses def_1, def_2;

def_2 is primary

uses def_2, def_1;

def_1 is primary

https://wiki.freepascal.org/Uses

Quote

However, two or more units listed in the uses clause might declare the same identifier. Then, only the identifier declared by the unit later in the list can be referred to using the short notation. Identifiers declared by earlier units in the uses clause can only be accessed via fully-qualified identifiers.

uses dd,crt;

begin
	writeln('COLOR_GREEN ', dd.COLOR_GREEN);
	writeln('GREY ', dd.GREY);
	readkey;
end.

Edited February 21 by tebe

roeoender · February 21

50 minutes ago, tebe said:
USES has recently been improved so that it works as in FPC
uses def_1, def_2;
def_2 is primary
uses def_2, def_1;
def_1 is primary

https://wiki.freepascal.org/Uses
uses dd,crt;

begin
	writeln('COLOR_GREEN ', dd.COLOR_GREEN);
	writeln('GREY ', dd.GREY);
	readkey;
end.

Thank you for explaination.

Prefixing with unit name is a good solution.

Hope this info will be helpful to others.

Your MP and MADS are great tools!

neuling · February 22

Hi good afternoon.
I play with the madpascal and the plus4 (Yape)

What does the memory control/memory structure look like?

I would like to start the program at $8000 so that I can use the storage underneath myself.

how can I do that please?

Thanks.

tebe · February 22

6 hours ago, neuling said:

I would like to start the program at $8000 so that I can use the storage underneath myself.

mp.exe file.pas -t c4p -code:8000

mads file.a65 -xi:<MadsPath>\base

neuling · February 22

mp.exe hello.pas -t c4p -code:8000

Hello thanks for the information.

But I can't find the start address in hello.a65: 8000?

Thanks.

----------------------------------

; ------------------------------------------------------------
; Mad Pascal Compiler version 1.7.1 [2024/02/19] for 6502
; ------------------------------------------------------------

STACKWIDTH = 16

TRUE = 1
FALSE = 0

; ------------------------------------------------------------

org $02

zpage

.if .def(@AllocMem)||.def(MAIN.SYSTEM.GETMEM)||.def(MAIN.SYSTEM.FREEMEM)
psptr .ds 2 ; PROGRAMSTACK Pointer
.fi

bp .ds 2
bp2 .ds 2

eax   .ds 4                       ;8 bytes (aex + edx) -> divREAL
edx   .ds 4
ecx   .ds 4

TMP
ztmp
ztmp8   .ds 1
ztmp9   .ds 1
ztmp10   .ds 1
ztmp11   .ds 1

STACKORIGIN .ds STACKWIDTH*4
zpend

; ------------------------------------------------------------

ax   = eax
al   = eax
ah   = eax+1

cx   = ecx
cl   = ecx
ch   = ecx+1

dx   = edx
dl   = edx
dh   = edx+1

org eax

FP1MAN0   .ds 1
FP1MAN1   .ds 1
FP1MAN2   .ds 1
FP1MAN3   .ds 1

org ztmp8

FP1SGN .ds 1
FP1EXP .ds 1

org edx

FP2MAN0   .ds 1
FP2MAN1   .ds 1
FP2MAN2   .ds 1
FP2MAN3   .ds 1

org ztmp10

FP2SGN .ds 1
FP2EXP .ds 1

org ecx

FPMAN0   .ds 1
FPMAN1   .ds 1
FPMAN2   .ds 1
FPMAN3   .ds 1

org bp2

FPSGN .ds 1
FPEXP .ds 1

; ------------------------------------------------------------
   opt h-f+
   org $1001
   org [a($1001)],$1001
   basic_start(START)

org $100E

CODEORIGIN

STATICDATA

; ------------------------------------------------------------

RTLIB
icl 'rtl6502_c4p.asm'

.print 'ZPAGE: ',zpage,'..',zpend-1

.print 'RTBUF: ',@buf,'..',@buf+255

.print 'RTLIB: ',RTLIB,'..',*-1

; ------------------------------------------------------------

START
tsx
stx MAIN.@halt+1

VLEN = VARDATASIZE-VARINITSIZE
VADR = DATAORIGIN+VARINITSIZE

   ift (VLEN > 0) && (VLEN <= 256)
   ldx #256-VLEN
   lda #$00
   sta:rne VADR+VLEN-256,x+
   eli VLEN>256
   m@init
   eif

.ifdef psptr
mwa #PROGRAMSTACK psptr
.fi

lda #$0e
jsr $ffd2 ; shifted charset

ldx #$00
stx bp ; BP = 0

dex ; X = 255

UNITINITIALIZATION

.local MAIN ; PROCEDURE

jmp l_0012

; ------------------------------------------------------------

.local SYSTEM ; UNIT

; ------------------------------------------------------------

M_PI_2   = $0648
D_PI_2   = $0192
D_PI_180   = $04
FMOPENREAD   = $04
FMOPENWRITE   = $08
FMOPENAPPEND   = $09
FMOPENREADWRITE   = $0C
SCREENWIDTH   = DATAORIGIN+$0000
SCREENHEIGHT   = DATAORIGIN+$0002
DATESEPARATOR   = DATAORIGIN+$0004
FILEMODE   = DATAORIGIN+$0005
GRAPHMODE   = DATAORIGIN+$0006
IORESULT   = DATAORIGIN+$0007
EOLN   = DATAORIGIN+$0008
RNDSEED   = DATAORIGIN+$0009

.endl ; UNIT SYSTEM

; ------------------------------------------------------------

.local CRT ; UNIT

; ------------------------------------------------------------

BW40   = $00
CO40   = $01
BW80   = $02
CO80   = $03
MONO   = $07
C40   = $01
C80   = $03
BLACK   = $00
WHITE   = $01
RED   = $02
CYAN   = $03
PURPLE   = $04
GREEN   = $05
BLUE   = $06
YELLOW   = $07
ORANGE   = $08
BROWN   = $09
YELLOW_GREEN   = $0A
PINK   = $0B
BLUE_GREEN   = $0C
LIGHT_BLUE   = $0D
DARK_BLUE   = $0E
LIGHT_GREEN   = $0F
BLINK   = $80

.endl ; UNIT CRT
l_0012

; optimize FAIL ('@print', hello.pas), line = 5

@printSTRING #CODEORIGIN+$0000
@printEOL
@exit

@halt ldx #$00
txs

rts

; ------------------------------------------------------------

.local @DEFINES
C4P
.endl

.local @RESOURCE
.endl

.endl ; MAIN

; ------------------------------------------------------------
; ------------------------------------------------------------

.macro UNITINITIALIZATION

   .ifdef MAIN.SYSTEM.@UnitInit
   jsr MAIN.SYSTEM.@UnitInit
   .fi

   .ifdef MAIN.CRT.@UnitInit
   jsr MAIN.CRT.@UnitInit
   .fi
.endm

; ------------------------------------------------------------

   ift .SIZEOF(MAIN.SYSTEM) > 0
   .print 'SYSTEM: ',MAIN.SYSTEM,'..',MAIN.SYSTEM+.SIZEOF(MAIN.SYSTEM)-1
   eif

   ift .SIZEOF(MAIN.CRT) > 0
   .print 'CRT: ',MAIN.CRT,'..',MAIN.CRT+.SIZEOF(MAIN.CRT)-1
   eif

.nowarn .print 'CODE: ',CODEORIGIN,'..',MAIN.@RESOURCE-1

; ------------------------------------------------------------

   ?adr = *
   ift (?adr < ?old_adr) && (?old_adr - ?adr < $120)
   ?adr = ?old_adr
   eif

org ?adr
?old_adr = *

DATAORIGIN
.by $28 $00 $18 $00 $2D $0C

VARINITSIZE = *-DATAORIGIN
VARDATASIZE = 11

PROGRAMSTACK = DATAORIGIN+VARDATASIZE

.print 'DATA: ',DATAORIGIN,'..',PROGRAMSTACK

; ------------------------------------------------------------

.macro STATICDATA
.by $0D $48 $65 $6C $6C $6F $20 $57 $6F $72 $6C $64 $20 $21
.endm

end

-------------------------------------------------------

zbyti · February 22

@neuling first of all I admire your MP tests but if I were you I choose RAW target for Commodore machines for $8000

Edited February 22 by zbyti

zbyti · February 22

@neuling and I'm really think you are @funkheld

Gury · February 23

It came to my mind also 😃

I suggest to neuling to open new topic for questions about Mad Pascal on its own.

neuling · February 29

I lost the password at Funkheld when I was 75 years old.

greeting

Gury · February 29

Great to see you switched to Mad Pascal👍

neuling · February 29

Hi, Thank You.
At 75 years old, I can no longer remember things that are more than 2 weeks old, that's why I ask so many questions.

I have another problem with starting a program life.pas for the X16 under Windows 11.

This is my request in the madpascal folder:
mp.exe -t x16 c:\madpascal\samples\x16\CRT\life.pas

but there is a message saying that it was entered incorrectly.

Thanks.
greeting

Edited February 29 by neuling

tebe · February 29

2 hours ago, neuling said:

with starting a program life.pas for the X16

MadRafi tries to add target x16, currently such a target does not exist.

target: a8, c64, c4p, neo, raw

neuling · February 29

this is in the madpascal.

https://github.com/MADRAFi/Mad-Pascal/tree/x16

greeting

MADRAFi · February 29

this is my fork. It is unfinished and probably never will be.

in order to use new target you need to compile mp.pas to mp.exe. Those in the repo are old ones and are not recompiled.

fpc -MDelphi -vh -O3 mp.pas

Edited February 29 by MADRAFi

neuling · February 29

Hello, thanks for the help.

will recompile it.

Your work for the x16 is wonderful.

please continue.

Thanks.
greeting

amarok · March 4

@tebe,

I am not sure, but I think I've just found a problem with memory alignment for variables with fixed address (set by absolute) which are types of records containing arrays. Please see the following source code:

program Test;

const
  SIZE = 3;

type
  TStructure = record
    array_of_bytes: array[0..SIZE - 1] of Byte;
    array_of_words: array[0..SIZE - 1] of Word;
    byte_field: Byte;
  end;

var
  auto: TStructure;
  fixed: TStructure absolute $4000;

begin
end.

There are two variables of the same type - auto, which address is automatically set by compiler and fixed with address set by hand.

When I look on lst file I can see that alignemnt for members of auto variable is proper but for fixed not. Please see the following excerpt from lst file.

   262 = 205F			adr.AUTO	= DATAORIGIN+$000B	; [10] RECORD
   263 2032			.var AUTO	= adr.AUTO .word
   264 = 205F			adr.AUTO.ARRAY_OF_BYTES	= [DATAORIGIN+$000B] .array [3]
   265 2032			.var AUTO.ARRAY_OF_BYTES	= adr.AUTO.ARRAY_OF_BYTES .word
   266 = 2062			adr.AUTO.ARRAY_OF_WORDS	= [DATAORIGIN+$000E] .array [3] .word
   267 2032			.var AUTO.ARRAY_OF_WORDS	= adr.AUTO.ARRAY_OF_WORDS .word
   268 = 2068			AUTO.BYTE_FIELD	= DATAORIGIN+$0014
   269 = 4000			adr.FIXED	= $4000
   270 2032			.var FIXED	= adr.FIXED .word
   271 = 4000			adr.FIXED.ARRAY_OF_BYTES	= $4000
   272 2032			.var FIXED.ARRAY_OF_BYTES	= adr.FIXED.ARRAY_OF_BYTES .word
   273 = 4002			adr.FIXED.ARRAY_OF_WORDS	= $4002
   274 2032			.var FIXED.ARRAY_OF_WORDS	= adr.FIXED.ARRAY_OF_WORDS .word
   275 = 4004			FIXED.BYTE_FIELD	= $4004

You can see that the local offsets of members are proper in case of auto variable:

- array_of_bytes: 0=$205F-$205F

- array_of_words: 3=$2062-$205F

- byte_field: 9=$2068-$205F

But local offsets of members are invalid for fixed variable:

- array_of_bytes: 0=$4000-$4000

- array_of_words: 2=$4002-$4000

- byte_field: 4=$4004-$4000

I am using the latest version of MadPascal and MadAssembler taken from github. Have you ever found the problem?

I really need to use records with fixed addresses for cartridge version of Time Wizard. Could you please fix it?

Thank you in advance for your help.

tebe · March 4

17 hours ago, amarok said:

I am not sure, but I think I've just found a problem with memory alignment for variables with fixed address

fixed

amarok · March 4

Many thanks, @tebe. You are really fast. I've checked the latest version and it works perfectly.

tebe · April 4

unit PP (Power Packer decompressor)

example

https://github.com/tebe6502/Mad-Pascal/tree/master/samples/a8/compression/pp

PowerPacker (gcc main.c)

https://github.com/lab313ru/powerpacker_src

Alfred · April 5

Just trying to build the sieve.pas example. MP compiles it fine, but mads has issues:

What file has the missing definitions ?

tebe · April 5

1 hour ago, Alfred said:

but mads has issues:

mads.exe filename.a65 -x -i:<MadPascalPath>\base

switch -x is important

Mad Pascal

Recommended Posts

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Join the conversation

Recently Browsing 0 members