Apple IIe - RAMWorks compatibility issue

[inexorabletash]

Try this on an Apple IIe with RAMWorks or compatible:

Boot to a BASIC prompt, then enter:

CALL -151
100: 20 00 C3 8D 00 C0 A9 32
108: 8D 00 04 A9 31 8D 05 C0
110: 8D 00 04 A9 01 8D 73 C0
118: A9 39 8D 00 04 4C 1D 01
100G

• On the composite output, the upper right should display 12 (inverse)
• On the VGA output, the upper right displays: 92 (inverse)

What's happening:

• hit 80STOREOFF (so PAGE2 doesn't have special meaning)
• write digit 2 to $400 main
• hit WRCARDRAM
• write digit 1 to $400 aux
• write 1 to $C073 to select RAMWorks aux bank 1 instead of 0
• write digit 9 to $400 aux

Analysis:

• Writing to RAMWorks banks other than bank 0 should not affect the 80 column display. I didn't spot any code in the project to handle RAMWorks banks.
• All of $C071, $C073, $C075 and $C077 should behave the same way. discussion
• Perfect behavior will be a challenge:
  • On a system without RAMWorks memory, $C073 would be ignored. So a program that writes to $C073 followed by writes intended for the 80col display would not be seen. However, such a program would behave incorrectly on a system with RAMWorks, and be considered buggy.
  • On a system with RAMWorks memory, selecting a bank with $C073 that does not exist will cause writes to go to another bank. Sometimes this would update bank 0, depending on the aliasing. Predicting this behavior would be a challenge if not impossible. Again, this would only occur in a buggy program.

Hardware: Enhanced Apple IIe, w/ A2VidGA, RAMWorks compatible from GW

[inexorabletash]

Here's a practical example, and how I noticed the problem in the first place:

1. Boot a ProDOS disk with an appropriate RAMWorks disk driver on it, e.g. John Brook's RAMAUX.SYSTEM which creates /RAMA
2. Without restarting ProDOS, launch Apple II DeskTop (from https://github.com/a2stuff/a2d/releases)
3. If A2D doesn't start copying itself to RAMDisk, go to Apple > Control Panels > Options and enable the option, then File > Quit and restart

What I see via VGA:

What I see via Composite:

Note that in the VGA screenshot, in the even columns at the middle of the screen you can see the inverse ...@@@@@? - this is the ProDOS volume bitmap slowly being populated as the file copy proceeds. The bitmap is apparently in Bank 1 between $400 and $7FF, with inverse @ being $00 or %00000000 which represents 8 in-use blocks, and inverse ? is $3F or %00111111 which is 2 used blocks and 6 free blocks.