KSCAN Examples

[GDMike]

I have seen so many KSCAN examples but im still having troubles. Sometimes the keys completely become useless and even a reboot doesnt fix it, but actually resetting the computer to run a program that just ran fine with KSCAN but NOT the second time back to back.

ive tried clearing 8374,5 837C on exit, on start..no matter what i try i seem to always have an issue, but not all the time but more than 60 perc of the time.

im usually going to write to the screen straight from a key routine..ive tried saving my R11 on BL routines and that isnt affecting my outcome. ive tried straight thru programming with no BL..still have the same issues. im using a NANO not sure why it would matter..But if someone could post a routine that works for them so i can test against it id appreciate it. thanks, here is a photo of my last kscan use...BTW spaceb was a Data >2000

 

[HOME AUTOMATION]

KEYMOD EQU  >8374
KEYVAL EQU  >8375
STATUS EQU  >837C
NOKEY  BYTE >FF
ANYKEY BYTE >20
*
*
*
SCAN   CLR  @KEYMOD
       LIMI 2
       LIMI 0
       CLR  @STATUS
       BLWP @KSCAN
       CB   @NOKEY,@KEYVAL
       JEQ  SCAN
       RT

* or

SCAN   CLR  @KEYMOD
       LIMI 2
       LIMI 0
       CLR  @STATUS
       BLWP @KSCAN
       CB   @ANYKEY,@STATUS
       JNE  SCAN
       RT
       
* call using

       BL   @SCAN

Edited July 2, 2019 by HOME AUTOMATION
I had listed the same line twice.

[GDMike]

thank you..a simple thing has turned into a three hr ordeal..yesterday i accomplished a huge amount of progress, but it all got smashed by a bad kscan. some illustrations say put a 5 into 8374 some say clr status and drop out..but  from what i know is that i want it to be a subroutine, i want to be able have different branches for the same key for diff areas within my prog. and i can accomp that by using an index type reg. so i can handle that. i cant wait to try this in its form. ty very much! i wasnt getting anywhere with the crap i had going on.

[+TheBF]

I remember doing a lot of trial and error to get a version than ran in Forth to my liking.

Here is my latest which test for a keystroke and moves it on the top of the Forth stack (cached in R4 here, called TOS) if a key-press is detected.

 

It shows how to use BL to call KSCAN by manually changing work-spaces to GPL and back to Forth. (In case it might be useful to someone)

Like FBForth, Camel99 runs with interrupts on so I only disable them while in the KSCAN routine.

 

Put your Yoda ears on when you read it and it makes more sense. ?

 

FYI:  The NE IF, line assembles the instruction   JEQ ENDIF

 

Edit:

I know about my terrible heresy, reading >8374 with MOV instead of MOVB but it works and I don't need to SWPB to put the character on the correct side for Forth to work with it

 

CODE: KEY? ( -- ?)  \ *WARNING* it takes 1,128uS for KEY? scan to run
            TOS PUSH,
            TOS CLR,            \ TOS will be our true/false flag
            0 LIMI,
            83E0 LWPI,          \ switch to GPL workspace
            000E @@ BL,         \ call ROM keyboard scanning routine
            8300 LWPI,          \ return to Forth's workspace , interrupts are restored
            2 LIMI,
            837C @@ R0 MOVB,    \ read GPL status byte (=2000 if key pressed)
            NE IF,
                8374 @@ TOS MOV, \ read the key
            ENDIF,
            NEXT,               \ return
            END-CODE

 

Edited July 2, 2019 by TheBF
Clarification

[+Lee Stewart]

45 minutes ago, TheBF said:

I know about my terrible heresy, reading >8374 with MOV instead of MOVB but it works and I don't need to SWPB to put the character on the correct side for Forth to work with it

 

Not that it matters here, but I do not think this will work for >8374 set to keyboard = 1 or 2 because >8374 is only reset to 0 by the console ROM’s keyboard scanner following setting >8374 to keyboard = 3, 4, or 5 and, of course, untouched if set to keyboard = 0 (default keyboard indicated at >83C6).

 

...lee

[+TheBF]

Thanks Lee. That's good to know.

[+TheBF]

50 minutes ago, Lee Stewart said:

 

Not that it matters here, but I do not think this will work for >8374 set to keyboard = 1 or 2 because >8374 is only reset to 0 by the console ROM’s keyboard scanner following setting >8374 to keyboard = 3, 4, or 5 and, of course, untouched if set to keyboard = 0 (default keyboard indicated at >83C6).

 

...lee

I only read 8374  (16 bits) when the status byte is non zero so I don't think I need to care if >8374 is reset to zero by the system?

 

A bigger question is that I just plugged a bunch of different byte values into >83C6 and I don't see a difference in my keyboard values.

Edit: I can see a difference.

Maybe I have to do it  every time I call KSCAN to get it to change. (?)

 

Ah  yes when I did this 

: KEYTEST   1 83D6 C!  KEY ;

I got difference values for control keys. So I will do some further testing to understand if my heresy can stand the inquisitional rigor of Atariage. ?

 

And still I am learning 

Edited July 2, 2019 by TheBF

[GDMike]

very interesting. ill be trying something out later today. thx so much for sharing.

[+mizapf]

After learning more about the TMS9900 architecture, I somewhat started to dislike MOVB and the other byte commands. MOV is OK, as long as you what what you are doing.

 

[+Lee Stewart]

4 hours ago, TheBF said:

I only read 8374  (16 bits) when the status byte is non zero so I don't think I need to care if >8374 is reset to zero by the system?

 

A bigger question is that I just plugged a bunch of different byte values into >83C6 and I don't see a difference in my keyboard values.

Edit: I can see a difference.

Maybe I have to do it  every time I call KSCAN to get it to change. (?)

 

Ah  yes when I did this 

: KEYTEST   1 83D6 C!  KEY ;

I got difference values for control keys. So I will do some further testing to understand if my heresy can stand the inquisitional rigor of Atariage. ?

 

And still I am learning 

 

Re >83C6:  Unless you never plan to use the console ROM’s KSCAN, I would not ever set the byte at >83C6.  That is for KSCAN to keep track of the “default” keyboard. It only gets changed when KSCAN detects a 3, 4 or 5 in the byte at >8374. When that happens, KSCAN effectively subtracts 3 from it, stores that value at >83C6 for future keyboard restoration to this new default and zeroes >8374. The next time KSCAN is run and detects 0 at >8374, it checks >83C6 to set the default keyboard to 3 (standard TI-99/4 keyboard scan), 4 (Pascal keyboard scan) or 5 (Basic keyboard scan).

 

Re reading word at >8374 to read new character into LSB in one operation:  The status byte’s COND bit is set for keyboards 1 (left half) and 2 (right half) when a key is struck, just as it is for keyboards 3, 4 and 5. But for keyboards 1 and 2, you will have a value greater than >FF for your new character value you store in TOS, viz., >01xx or >02xx, not >00xx as you expect because the byte at >8374 does not get zeroed. Your ruse only works for full keyboard scans (0 [default], 3, 4 and 5).

 

...lee

[+TheBF]

I am laughing out loud regarding your use of the word "ruse".?

Excellent. If the shoe fits I guess.

 

Thanks as always for the extremely lucid description.

What with Flat earth advocates and Moon Landing deniers I may have to re-consider my options lest my ruse be lumped together with that lot. ?

 

0xBF

[GDMike]

On 7/2/2019 at 12:43 AM, GDMike said:

thank you..a simple thing has turned into a three hr ordeal..yesterday i accomplished a huge amount of progress, but it all got smashed by a bad kscan. some illustrations say put a 5 into 8374 some say clr status and drop out..but  from what i know is that i want it to be a subroutine, i want to be able have different branches for the same key for diff areas within my prog. and i can accomp that by using an index type reg. so i can handle that. i cant wait to try this in its form. ty very much! i wasnt getting anywhere with the crap i had going on.

i was easily able to make this work, it was very similar to what i had been experimenting with..and the key selection worked, but my next command, a BL @cls did not, or sorta kinda did..

i selected key 1 to take me to a cls and only part of the screen cleared..apparently it does not like my mov R11,@Savrtn command because i got rid of my BL @cls and just went straight to a clear screen..too bad..but apparently it was NOT my kscan all along..but my returning R11..#$#@

----

SAVRTN DATA 0

...

then i start a BL routine with:

mov R11,@savrtn

and end with

mov @Savrtn,R11

RT  *R11

im sure i tried B *R11 as well but i could try it again...and 5 mins. later, none of those worked..well ay least i now know why half of my sub programs arent working, as my sound routine works once and no more after calling it as well. so until i can wrap my head around the R11 rt, or RTWP or B *R11 ill just write straight thru code.

 

Edited July 3, 2019 by GDMike

[+FarmerPotato]

Do you have a LWPI in your program?

 

Normally you don't need one because when started from E/A your program has WP = 20BA

Essentially "Load and Run" has already done this for you:

   LWPI >20BA

For instance you could do,

MYWS  BSS 32

START  LWPI MYWS
       LIMI 0

This will make your program's registers reside in MYWS which you own.     It's a good practice, to have your program to set up things for itself. (It might be launched from DM1000 for example)

 

What you should never have is LWPI >83E0  or LWPI GPLWS  if you are going to use KSCAN. They will fight.

 

 

Here is the source to the E/A version of KSCAN. It actually is polite enough to save R11, though it will .

GPLWS  EQU >83E0
UTLWS  EQU >2094
KSCAN0 EQU >21DE
KSCAN  DATA UTLWS,KSCAN0

KSCAN0
   LWPI GPLWS              Set WS = GPLWS
   MOV  R11,@UTLWS+22      Save the GPLWS R11 by putting it in our R11, haha
   BL   @SCAN              BL uses R11 of GPLWS
   LWPI UTLWS              Set WS = our WS
   MOV  R11,@GPLWS+22      Restore GPL's R11              
   RTWP  

 

 

 

[GDMike]

i have this

[GDMike]

ok, now it works...why..let me take a look at why..oh i added LIMI 0

earlier i had Limi 2 limi 0

so after i couldnt get a change i then removed it altogether and got further than i was getting with half a screen getting cleared...lol

i had a routine that included a sound table and it would always work once but never twice, so i found a GPLlnk routine to make a beep.. (since im hitting numbers at a menu i wanted a beep).

and the GPLlnk routine always worked. now maybe i can try my sound table all over again.

i call this progress...but one day ill figure out my addresses that perform those basic functions without the use of Defs..

i know thats the best route, and i use to try to include those, but as soon as i started losing WP info i went back to the basics to attempt to get things in a staggering forward direction as to where i was, a drunkard Hobo looking for a bottle..

 

Edited July 4, 2019 by GDMike

[GDMike]

thank you all so much!!! im making lots of progress on my little routine and its never run so smoothly! i dont even have to make more than one Kscan type of subr. everything i BL to actually comes back..im thrilled.

Edited July 4, 2019 by GDMike