> Here's a project I'm working on, eventually it'll be used on an
 > old PC (8088) serving as a dumb terminal via null modem cable
 > into a Linux box. It has a problem with the escape sequences used
 > for the display, [;80H for example fails miserably. These
 > are in the procedure called "EscSquare"- a real headache.
 > begin 644 term.zip
[clipped]
 > end
"A real headache" indeed. I hope you didn't write the layout of the code 
yourself (or you don't mind me criticising it ;-) ) ...
I took just a 15 minute glance of the code and didn't directly find anything 
wrong with the EscSquare routine, but I would rip out most of the code and 
replace it with data structures as follows (all code typed right into the 
message, typos included; Numerics should be OK since copied from Pascal code 
;-) ) :
Flexibilize section :
VidMem      dw      ?           ; You'll love this when you find an old
                                ; Hercules mono adapter...
ScreenCols  dw      ?           ; Or maybe you find a VGA and want to
ScreenLines dw      ?           ; use some weird screen resolution ...
ScreenSize  dw      ?           ; Of course all of these must be set          
                       ; up...
biosSeg             equ         0040h
biosDisplaymode     equ         049h
biosDisplayColumns  equ         04Ah
biosDisplayRows     equ         084h    ; Number of rows -1 (on EGA+)
biosDisplaySize     equ         04Ch    ; Total number of bytes per           
                              ; screen
vidBaseColor        equ         0B800h
vidBaseMono         equ         0B000h
InitializeScreen proc near
                    push    es
                    push    biosSeg     ; Point es to the BIOS                
                         ; parameters
                    pop     es
                    mov     ax, vidBaseColor
                    cmp     byte ptr es:[biosDisplayMode], 07h
                    jne     @@1
                    mov     ax, vidBaseMono
@@1:
                    mov     [VidBase], ax
                    xor     ax, ax
                    mov     al, es:[biosDisplayColumns]
                    mov     [ScreenCols], ax
                    mov     al, es:[biosDisplayLines]
                    inc     ax
                    mov     [ScreenRows], ax
                    pop     es
                    ret
endp
For reading a char from the buffer, I would use _one_ centralized routine, 
that waits depending on (eg) the carry flag, and returns with the carry flag 
set, if a char is returned, no carry->no char available

I would rip out the crude jump table in the routines Control and Escape
and replace them with the following table/code :
ControlTable    label   word
        dw      offset notImplemented       ; NULL
        dw      offset notImplemented       ; Start of heading, ^A
        .
        .
        .
        dw      offset Escape               ; Escape  ^[
Control proc near
; AL = control char just read, AL < 32 !
            cbw                             ; AX <- AL
            mov     bx, ax                  ; BX = AL
            add     bx, ax                  ; BX = 2*AL
            call    ControlTable[bx]
            ; You could even jmp ControlTable[bx] ...
            ret
endp
The carriage return routine is a mess. Here is my try at a better one that 
works with every screen resolution :
CarriageReturn proc near
; Positions DI to the beginning of the next line, scrolling not up if ; 
necessary.
            mov     bx, ScreenCols
            add     bx, bx              ; Number of bytes between every
                                        ; first column
            mov     cx, ScreenSize      ; CX = first column of last line
            sub     cx, bx
            xor     ax, ax              ; Try with (0,0)
@NextLine:
            cmp     ax, di              ; current line below cursor           
                              ; position ?
            ja      @Found
            add     ax, bx              ; Position one line down
            jmp     @NextLine           ; and try again
@Found:
            mov     di, ax              ; accept new cursor position
            cmp     di, cx              ; And fix position if beyond          
                               ; last line on screen
            jbe     @Done
            mov     di, cx              ; fix cursor position
@Done:
            ret
endp
Escape itself is a horror too. Here I would also replace it directly with 
offsets of the procedures to call ...
I hope I've given you enough ideas to restructure the whole thing, and maybe 
while rewriting, the bugs disappear ;-)
-max
---
---------------