Path: news.uiowa.edu!chi-news.cic.net!vixen.cso.uiuc.edu!uwm.edu!math.ohio-state.edu!pacific.mps.ohio-state.edu!freenet.columbus.oh.us!not-for-mail
From: dalloff@freenet.columbus.oh.us (Dave Althoff)
Newsgroups: comp.sys.apple2,comp.sys.apple2.programmer
Subject: 80-col. firmware bug: help!
Date: 10 Mar 1996 21:24:45 -0500
Organization: The Greater Columbus FreeNet
Lines: 62
Message-ID: <4i02td$q44@acme.freenet.columbus.oh.us>
NNTP-Posting-Host: acme.freenet.columbus.oh.us
Summary: Someone tell me about a long-documented bug...
X-Newsreader: TIN [version 1.2 PL2]
Xref: news.uiowa.edu comp.sys.apple2:100313 comp.sys.apple2.programmer:6381

It's not a new bug, this one, I think, is documented...

I was writing a text file display program for my ][e, which will
ultimately become part of a file-browsing program for a particularly
specialized application in the future, when I ran into a bug in the
80-column firmware.  I'm writing and testing this program on my ROM 01
IIgs, since it runs ProDOS, and I am opposed to running that miserable OS
off of floppies on the ][e.

(Yes, ProDOS is miserable, particularly when you're floppy based and
talking to it from BASIC.SYSTEM.  But that's another issue.  The project
will ultimately use ProDOS, but it will also run off of a 20Mb Vulcan).

Anyway, here's the problem.  I use the following to write text to the screen:

30 SHOLINE    LDY #$00
31            STX $25       ;POKE 37,X-reg.; X is pre-loaded
32            JSR VTAB      ;Recalculate BASL/BASH
33 SHOLP1     LDA LINE,Y    ;Get the next character
34            STY $24       ;POKE 36,Y-reg
35            JSR COUT      ;Show the character
36            INY           ;Advance
37            CPY WIDTH     ;Done with this line?
38            BCC SHOLP1    ;No: Do next character

(etc).

Here's the problem.  It all works great, except that the cursor is not
repositioned to the 0th column on line 34 when Y=0.  For all other (legal)
values of Y, the routine works just fine.  But putting a #$00 into $24 has
no effect on the cursor position.  

This is, I think, a documented bug, and it exists on everything from the
original ][e on up.  I thought it might have been squashed in the enhanced
//e, but its appearance on my IIgs indicates that is not the case.

I vaguely remembered from typing in the "MAGIC.MENU" program many years
ago (anyone remember that one?  It was in the ProDOS Applesoft
Programmer's Tool Kit) that there was some junk about comparing the value
in PEEK(36) with some other location then POKEing location 1403.  So I
added between lines 34 and 35:
35           STY $057B   ;Squash 80-column bug

And now the program works just fine.

SO...Can anyone give me a little background on this bug?  Will I break
something by ALWAYS writing the horizontal cursor position to that screen
hole at $057B?  Or do I need to remember that other location and do that
compare?  Is this in a technote somewhere?  Can I find it on the 'Net
somewhere?

NOTE:  My server doesn't carry c.s.a2.programmer, so if you're reading
this there (I'm exploiting a "bug" in Usenet here), please include c.s.a2
in any followups.  Thanks!

--Dave Althoff, ][

-- 
    /-\        _       _               __*** Closed for the season ***
   /XXX\      /X\     /X\_      _     /XX\_      _     _        _____  
  /XXXXX\    /XXX\  _/XXXX\_   /X\   /XXXXX\    /X\   /X\      /XXXXX
_/XXXXXXX\__/XXXXX\/XXXXXXXX\_/XXX\_/XXXXXXX\__/XXX\_/XXX\_/\_/XXXXXX