Well, I've been hacking on my Kestrel-2 emulated environment again. I'm finding the Forth firmware image to be extremely buggy. The bugs didn't manifest themselves until fairly recently, as I was attempting to write Forth storage drivers.

Oh, I also have implemented the first (emulated) SerBus peripheral for the Kestrel, too -- a 64MB capacity direct access storage device (DASD) with 1024-byte 'sectors.' In other words, it's purpose-built for Forth.

However, I'm finding that the Forth image in ROM has entirely too many bugs for my liking, so I'm planning on starting the Forth environment over yet again. I've also been considering ditching Forth in favor of another language, such as a small Lisp environment. An APL-inspired language is also not outside the question.

Currently, I'm working on some ROM-resident code to implement a nice graphical operating system. My current inspiration is GEOS/64, where I've just finished implementing the HorizontalLine clone. However, there are some differences from the GEOS code:

1) my code is MUCH faster than the GEOS code. It's completely 16-bit for starters, and also does a lot less work inside its inner loops.

2) my code supports drawing into off-screen bitmaps and even sub-bitmaps (a bitmap within a larger bitmap), which makes implementing course-grained windowing a cinch.

3) All parameters are passed on the stack, so there are no global direct-page locations used by the code.

UPDATE Below is a screenshot. Here is the code that was used to produce that screenshot:




This is generated by alternating between $AAAA and $5555 for the pattern when filling the whole screen. Then, I used $0000 to render the dropshadow, and finally $FFFF for the white index card.