Re: A89: kernel/no stub

To: <assembly-89@lists.ticalc.org>
Subject: Re: A89: kernel/no stub
From: "Zeljko Juric" <zjuric@utic.net.ba>
Date: Wed, 16 Aug 2000 09:36:11 +0200
Reply-To: assembly-89@lists.ticalc.org

Hi!

> I'm reading the exact quote you refered to right now, in the 
> newest version of the lib. And I'm gonna have to ask if this
> was an error on Zeljko's part.

I saw using EV_hook for SHIFT+ON in TEOS source: in teos.asm
new EV_hook is installed, and in prog.asm hook handler is
implemented, see label newEV_hook.

> Doors was using shift+on loooong before the event-driven 
> interface calls were ever even known - as in _years_ before
> (hard to imagine the 89 has been out over 2 years now). It
> should be a simple hook into int 6, which is called whenever
> the on button is pressed - write an int 6 wrapper that does
> whatever if desired key is down when the code is run.

No. As I said, I saw EV_hook in TEOS, and I was shocked that
DoorsOS also uses EV_hook. Note that events are not invented
by me and discovered by me. What is unusual in the fact that
someone knew about it before I documented it.

> Furthermore, I'm not sure that EV_hook is even capable of 
> handling such an instance.

It seems that it is: see newEV_hook in TEOS source. If anybody
is curious, here is a paste from TEOS source:

newEV_hook:   movem.l d0-d7/a0-a6,-(a7)
              cmp.w #$710,(a2)           ;keypress?
              beq.s toucheON
              lea posSHIFT(pc),a0
              move.b CALC(pc),d0
              addq.b #1,d0
              btst d0,7(a2)
              beq.s NONshift
              clr.b (a0)
              bra.s fin
NONshift:     st.b (a0)
              bra.s fin
toucheON:     cmp.w #267,10(a2)          ;ON key?
              bne.s fin
              move.b posSHIFT(pc),d0
              bne.s fin
              lea shell(pc),a5
              bsr find                
              tst.l d0
              beq.s notfound
              jsr (a0)
              bra.s fin
notfound:     pea shellnotfound(pc)
              ROM_CALL ST_showHelp
              addq.l #4,a7
fin:          movem.l (a7)+,d0-d7/a0-a6
              rts
shell         dc.b "doors",0,0,0
shellnotfound dc.b "Shell not found",0
posSHIFT      dc.b 0

> While I'm not too familiar with the TIGCC lib implementation
> of the event handlers,

It is just call of TIOS event handling routines,

> I have read some extensive documents on using them the way TI
> intended, and the applications do not appear to receive any 
> message when an error is triggered - and the on button 
> triggers error: break, not CM_KEYPRESS - correct?

No, ON key triggers CM_KEYPRESS, as you can see from TEOS code
fragment. If TI gave the wrong informations, this will not be
the first time from them to give incorrect informations.

> Although, now that I think about it, doors may have changed
> shift-on managment with the recent rewrite, it would certainly
> be a waste not to use int 6 instead, and TEOS certainly prefers
> the int 6 method.

Just as you can see, TEOS certainly does not prefer the int 6
method (at least, not in release 1.02 which I analysed).

Cheers,

Zeljko Juric