<
aiia...@gmail.com> wrote in message
news:2078785.21.1337087316506.JavaMail.geo-discussion-forums@yneo7...
Lines 370-390 look like interrupts are being turned off and the character
ROM switched into visible memory for the purpose of copying a character
pattern into the free memory block, then the character ROM is switched back
out and interrupts re-engaged. No idea why, though.
All the PRINT ""; statements are bit puzzling since they don't appear to do
anything. I'll guess that the special (ie., non-ASCII) characters originally
there aren't showing up in the listing. They would most likely have been
"screen clear" and/or "cursor home", ie., the KERNEL character output
routine would have treated them specially.
As has been mentioned, all the POKES based on an address starting at 1024
are directly modifying screen memory, ie., the memory the video chip scans
looking for the character codes of what it should display. They're not
directly concerned with the logic of the neural net but only with how things
look on the screen.
The lines 1840 and 2030 form a command to access a sequential file on drive
0 for writing and reading, respectively. The OPEN statements one line 1850
and 2060 actually open the files as #5 (an arbitrary number) on device 8
(usually a 1541 floppy drive, which actually had only one physical drive,
but not all Commodore drives were so limited). The file #5 is used to refer
to those files in subsequent writes and reads.
Line 50 OPENs file #15 on device 8 with a secondary value of 15. That
secondary value is a special value that accesses the "command channel" of a
drive (as opposed to a "data channel"). It's common and convenient to use
the same number for the file number (but not required). This file is used in
lines 2230+ to ask the drive whether or not an error has occurred during a
write or read (the variables are essentially "error number", "error message
(ie., text)", "error track", "error sector").
- Anton Treuenfels