> Hello Denis,
 > I didn't like what Debug did with your int2etst.dbg (512
 > bytes), so I had A86 assemble it. 17 bytes. QEMM took
 > exception. Tried without QEMM and got your results.
 >     * Note to A86 users *
 > Added mov ss,cs and A86 gave no error but changed the code
 > "behind my back"
 > to push cs  pop ss  which of course did not work!
 > (yes I know, mov ax,cs/mov ss,ax)
 > Here is what I sent to A86.
 > Mov     Bx,0200h        ;DOS:set memory block size
 > Mov     Cl,04           ; BX=paragraphs
 > Shr     Bx,Cl           ; ES=block seg
 > Inc     Bx              ; BX=para avail
 > Mov     Ah,4ah          ; ret if error
 > Int     21h
 > Mov     Si,0080h
 > Int     2eh             ;DOS Internal
 > nop
 > nop
 > mov     ss,cs       ;should be via ax.  A86 assembled this as
 > push cs/pop ss
 > mov     sp,0fffeh
 > mov     ah,4ch
 > int     21h
 > Kind regards, Barry
Don't'cha just _love_ Stone Age attempts at AI in an assembler?
I'd expect a compiler to do strange and wondrous things in attempts to 
optimize, do memory models, etc. etc.
But not a bleedin' assembler!  I wants my bytes where I puts 'em, and not 
what some bleedin' filthy assembler author decides is better for me .. or 
helps him enforce his licensing ...
And people wonder why A86 requires a religious conversion :-(
---
---------------