🚀 go-pugleaf

Since Bill has expressed curiosity about some of the details of my
AtariWriter Plus conversion to 80 columns, I will talk a little bit more
about it since it might be of general interest.

AW+ uses the extra memory of the 130XE (or compatible) in a kludgy sort of
way. (At least it appears to be kludgy; I have not used the word processor
enough to speak from experience.) Of the 5 16K banks that map into the
$4000-$7FFF range, the first 3 (the main bank and the first 2 extended
banks) are used as 3 independent 16K text buffers. The fourth is used as
the kill buffer and the fifth is largly unused: only the first 4K is used
for miscellaneous routines. That leaves 12K of this bank unused. HMMM.
Plenty of room for the 8K 80 column screen if only we could get Antic
to look up there.

Now Bill was wondering why I do not use the cartridge area for the 80 column
screen. The answer is that all that area has already been used by the
program. It is true that there is about 1K at the top of memory that is
used for the 40 column screen, but that is way too small for the 80 column
screen.

Back at the original problem, how do we get Antic to look way up into that
last extended bank? Pondering this question, I was suddenly struck by ATARI's
foresight (and Newell's for following the same convention in his 256KXL)
of giving us the control over the Antic to force it to stay with the main
memory bank even when the CPU is off looking at other banks. What if I were
to rearrange the order in which the banks are used so that the last bank
(the one with the free 12K) becomes the main bank? Then I could leave bit
5 of $D301 high, forcing Antic to look at my 80 column screen while the
CPU was off merrily flipping banks.

That is exactly what I have done. Fortunately, redefining the order of the
banks was easy; there are 5 little routines used to select each of the
5 banks. I merely swapped the bank address in the first and last routine.

Unfortunately, the rest of the conversion has been very tedious. In my
opinion, the program is quite poorly written. The screen output is not
at all elegant or structured. It is brute force and spread out in bits
and pieces throughout the code. However, the conversion is about 90%
complete and I hope to have it finished by June 1st. It will take the form
of a binary load file that will be appended to the AP.OBJ file on the
AtariWriter Plus disk.

David Young
-------