Mouse suggestions...

[576XE]

Hello friends! (ATARI LOVES YOU!)

This is the code converted from MAC/65 tokens to ATMASII source.

It's well known algorithmus for converting STM1-mouse GrayCode into PM movement commands and it's work.

IT USES:

TIMER for reading and deciphering Gray-Code to mouse data
PMG for drawing cursor
VBI for moving cursor

I'm not a games programmer. And I'm not a programmer at all!
I only wanted for about thousands of years to write mouse program for Atari!
Here is routine is known as very fast etc...

But I Personally think that The Native Atary - is TEXT interface.
Text but NOT Char-keyed!!!
(YES! I can imagine as acceptable Char-keyed interface in ESC-sequences combination key command-mode)

But really I need TEXT-Cursor Inverse mode for controlling Atari.

Here is code for getting pixel controlled cursor movement...

*
* ATARI  STM1  MOUSE Driver
*
* OS Equates
PMBASE EQU $D407
STRIG1 EQU $0285
PCOLR0 EQU $02C0
COLOR3 EQU $02C7
GPRIOR EQU $026F
PORTA  EQU $D300
SDMCTL EQU $022F
COLPM0 EQU $D012
GRACTL EQU $D01D
HPOSP0 EQU $D000
HPOSM0 EQU $D004
STRIG0 EQU $0284
NMIEN  EQU $D40E
VDSLST EQU $0200
SDLSTL EQU $0230
SDLSTH EQU $0231
VCOUNT EQU $D40B
DMACTL EQU $D400
*
SETVBV EQU $E45C
SYSVBV EQU $E45F
XITVBV EQU $E462
*
POKMSK EQU $10
VTIMR1 EQU $0210
IRQEN  EQU $D20E
AUDC1  EQU $D201
AUDF1  EQU $D200
STIMER EQU $D209
CRITIC EQU $42
*
P0     EQU   $0400
PMZONE EQU $8000
*
SHLEN EQU SHEND-SHAPE
*
       ORG $8000
*
* Init TIMER for reading mouse
INIT SEI
    LDA #MOUSE
    STA VTIMR1
    LDA #MOUSE/256
    STA VTIMR1+1
    CLI
    LDA POKMSK
    ORA #1
    STA POKMSK
    STA IRQEN
    LDA #0
    STA AUDC1
    LDA #192
    STA AUDF1
    STA STIMER
* Init PMG for drawing Arrow
    LDA #$3A
    STA SDMCTL
    LDA #$02
    STA GRACTL
    LDA #PMZONE/256
    STA PMBASE
CLEAR LDY #0
    TYA
CLR STA PMZONE+P0,Y
    INY
    BNE CLR
* Init VBI for moving cursor
    LDA #6
    LDX #CURSOR/256
    LDY #CURSOR
    JSR SETVBV
    RTS
*
* Read mouse in Timer
MOUSE TXA
    PHA
    TYA
    PHA
    SEI
    LDA CRITIC
    BNE SKIPIT
    LDA #0
    STA NMIEN
*
RDPORT LDA PORTA
    LSR
    LSR
    LSR
    LSR
    PHA
    AND #3
    ORA OLDX
    TAX
    AND #3
    ASL
    ASL
    STA OLDX
    LDY XPOS
    LDA MOVTAB,X
    BMI STORX
    BNE DECRX
INCRX INY
    CPY MAXX
    BCC STORX
DECRX DEY
    CPY MINX
    BCC INCRX
STORX STY XPOS
*
    PLA
    LSR
    LSR
    AND #3
    ORA OLDY
    TAX
    AND #3
    ASL
    ASL
    STA OLDY
    LDY YPOS
    LDA MOVTAB,X
    BMI STORY
    BNE DECRY
INCRY INY
    JMP CHECKY
DECRY DEY
CHECKY CPY MINY
    BCC INCRY
    CPY MAXY
    BCS DECRY
STORY STY YPOS
*
    LDA #$C0
    STA NMIEN
SKIPIT CLI
    PLA
    TAY
    PLA
    TAX
    PLA
    RTI
*
* Move Cursor in VBI
CURSOR LDA XPOS
    STA HPOSP0
*
    LDY #0
    TYA
CL0 STA PMZONE+P0,Y
    INY
    BNE CL0
*
    LDY YPOS
    LDX #SHLEN
MOV LDA SHAPE,X
    STA PMZONE+P0,Y
    INY
    DEX
    BPL MOV
*
    LDA STRIG1
    BNE NPRESD
    LDA #$0E
NPRESD STA PCOLR0
*
    JMP SYSVBV
*
OLDX DFB 126
OLDY DFB 128
XPOS DFB 126
YPOS DFB 128
MAXX DFB 208
MAXY DFB 224
MINX DFB 48
MINY DFB 32
*
* The Indextable 0=no action!
MOVTAB   DFB 255,1,0,255,0,255,255,1
         DFB 1,255,255,0,255,0,1,255
* Cursor shape
SHAPE    DFB $01,$12,$1C,$3C
         DFB $3E,$78,$60,$80
SHEND    EQU *-1

I ask you about CHAR-cursor CONTROLLED!!!

YES! I can CMP old value and new value of XCOORD and then make delay of drawing new position of TEXT cursor,,,

I think that there are hundreds desisions how to make it perfectly...

Really I want to save intact pixel-controlled cursor data (for further use) and add to program TextCursor-controlled data.

Your suggestions,please!!!

[Rybags]

You can't easily control the text cursor with a mouse. Simply changing the cursor X/Y doesn't actually get mirrored on the screen until the E: device produces output, ie when a key is pressed. Also having the cursor moved while program output is occurring would just create a huge mess.

 

What you could do though is have a mouse handler that translates mouse movement into keystrokes. Diagonal could be done by queuing 2 keystrokes.

You'd have to do the programming so that it doesn't clash with normal keyboard input.

 

In the end though I don't think it'd be very usable. For starters the OS can only process keystrokes at the rate of about 90 or so per second. And controlling a text cursor direct from mouse is kind of clunky. Lastly, the keystroke noise generated would be really annoying but at least on XL/XE you can turn the click off.

[576XE]

Hi there Rybags!

 

Please look at this code

...
LDA MOVTAB,X  ;Get table data which are 255 or 0 or 1
    BMI STORX ;When 255(or -1) NO ACTION! only store new X data
    BNE DECRX ;If not 0 then decrement cursor position
INCRX INY     ;If 0 then increment cursor position
...

My question is - "A me interpreted the code correctly?"

 

The documentation says that ZERO means NO ACTION TAKEN!

[flashjazzcat]

Yes, it's a look up table and the zero entries mean no movement on that axis.

[576XE]

Hi FJC.

 

May be I slightly skip some education...

 

I think that the code BMI STORX goes to ONLY storing HOR values and then works on Y values.

Thus this code skips all treating of input data concerning HORIZONTAL...

 

Brunch with MINUS says that we all - hand's up.

 

I can understand that being NOT a programmer I skip some schoolarships BUT?!!

 

I want to say that Minus skips all treatment.

[flashjazzcat]

I think that the code BMI STORX goes to ONLY storing HOR values and then works on Y values.

Thus this code skips all treating of input data concerning HORIZONTAL...

 

Brunch with MINUS says that we all - hand's up.

Hi!

 

Yes you're quite right. This code is some distant cousin of the mouse driver in the GOS, but I had forgotten just how many modifications were made by then. The GOS mouse driver simply adds the values in the LUT to the X and Y coordinates, so in that case, $FF = -1, 1 = +1, and 0 = no change. But the documentation for your routine seems at odds with what's actually happening.

 

The lookup table corresponds to the GOS lookup table, but with the sense of the values altered, so it should at least work.

 

This is the same section of the GOS mouse driver:

	.proc Mouse
	lda $d300

	cmp old_mouse
	bne Moved
	pla
	rti
Moved
	sta old_mouse
	and #$f0
	lsr ; not really worth using a LUT for this
	lsr
	lsr
	lsr
	pha
	stx mtmp1 ; save x

	and #3 ; ST mouse only
	
	ora oldx
	tax
	lda mousetab3,x
	sta oldx
	lda x_change
	clc
	adc mousetab,x
	sta x_change
	pla
	
	tax
	lda mousetab2,x
	
	ora oldy
	tax
	lda mousetab3,x
	sta oldy
	lda y_change
	clc
	adc mousetab,x
	sta y_change
	
	ldx mtmp1 ; restore x
	pla
	rti
	.endp
	
; ST mouse index table
	
mousetab ; 0 = no movement, 255 = -1, 1 = +1
	.byte 0,255,1,0,1,0,0,255,255,0,0,1,0,1,255,0

mousetab2
	.byte 0,0,0,0,1,1,1,1
	.byte 2,2,2,2,3,3,3,3
	
mousetab3
	.byte 0,4,8,12
	.byte 0,4,8,12
	.byte 0,4,8,12
	.byte 0,4,8,12

The VBI then adds x_change and y_change to the X and Y coordinates. This saves having to handle 16-bit values in the timer IRQ and software mouse acceleration ensures 8 bits are sufficient for one frame's worth of movement.

 

It's basically the same idea, but using signed direction changes in the lookup table.

[576XE]

I'm not skilled enough to read clearly this code, but I'll trying!

 

Thank you, FJC.

 

GOD BLESS YOU AND YOUR FAMILY AND CAT!

[576XE]

Hi, there FJC!

 

I have some questions about your code.

 

1. Really I think that you've used LUT in this:

Moved
    sta old_mouse
    and #$f0
    lsr ; not really worth using a LUT for this
    lsr
    lsr
    lsr

Because

Using AND #$F0 for clearing Least Nibble is not needed when we are clearing them with 4*LSR.

 

I really can not to clearly read all logic but I see that

 

- moustab can be renamed by movtab (it gives mouse deltas) or deltatab...

- moustab2 can be renamed as losntab - Low Semi-Nibble Tab (or HORTAB)

-mousetab 3 can be renamed as hisntab - Hi SemiNibble Tab (or VERTAV)

 

It's some reminense to Joystick values...

 

Here I ask about a logic of creating of ...

...

I got it!!! (Such a good boy!!!)

 

Anyway, this program is MUCH MORE clearer and elegance then some others.

 

It's only emulates own Not GrayCode above real GrayCode from mouse!!!

 

and returns one-step deltas.

 

(Really it's not a very distant cousine...)

 

One realized in assembly thinking manner - so unreadable as Basic

Other realized in human logic thus someone who reads something about Gray-code logic can see all picture.

 

And moving-deltas is a charm!

 

IMHO you said about "16-bits values" erroneously. (It's a 2-byte word...)

Your meaning was - some decimal value from 0 to 15 and thus - Least Nibble.

 

I still can not imagine where must be Delay counter if your code is in timer FE...

[flashjazzcat]

Yes, I think the cost of the AND plus the indexed table lookup was equal or greater to three rotates. Other names for the LUTs seem reasonable to me.

 

When talking about 16-bit numbers, I meant the actual coordinate system used by the mouse pointer. Actually only 16-bits on the X-axis, of course (since there are 320 pixels). So the signed 8-bit relative changes accumulated during the timer interrupt are added to X and Y every vertical blank.

 

Diamond GOS, incidentally, used a different scheme. Mouse was sampled using DLIs (spread across the display every 20 lines or so) and the NMI simply queued up raw mouse port readings which were then translated into movement in the VBI (IIRC).

Edited June 24, 2016 by flashjazzcat

[576XE]

My personal Best Wishes! Good old Yo!

[flashjazzcat]

To you also. Actually make that 9 bits for the X-axis on the GR. 8 screen, of course (and the Antic display portion of the driver capitalises on there only being 1 bit of the MSB used).

[pirx]

Diamond GOS, incidentally, used a different scheme. Mouse was sampled using DLIs (spread across the display every 20 lines or so) and the NMI simply queued up raw mouse port readings which were then translated into movement in the VBI (IIRC).

I've used the same trick in Operation Blood and other games with mouse support, but only 8 reads on DLI per frame (as many as charset changes).

[576XE]

You all may laugh or cry ...

 

But I'm in writing a letter which must clearly explain my daughter

(She is absolutely not a programmer THUS she understands LISP [really] clearer then me...)

How The Mouse works.

 

I've said her that Gray invented the scheme how to expand a rule and how to change it to staying Gray...

 

... mirroring -> demirroring ... Such a Good Boy!!! Anyway!

 

Are me right that in the case when we are not need position-controlled code like a Gray's we can simulate just binary coded tables?

I mean just use Demirroring schema...