Florence

Changes

• Finished port of IWM and threading code.

• Built initial VGC support.
• Name changed and versioning reset.

• Completed initial conversion of CPU, MMU and debugger to Objective-C under Cocoa. Now runs as an OS X native app. Not bad for only a week's work.

• Some more speed improvements (around 10%) for standard hires through some additional look up tables. (Lucky that memory is cheap these days)
• Speed improvements overall by around 20% through debug string optimizations.
• Hires wasn't displaying hi bit colours (blue and orange). Fixed.

• Replaced debugger button names with remote control style icons and tool tip text.

• Improved graphics performance by 4x with some double buffering and image caching tricks.
• Emulated code running on text page wasn't fetching 8 bit bytes, causing an exception. Fixed.
• New command line switch -jv to display Java version information on startup.
• Time stamp now displayed on console lines.

• New command line switch -meter to turn on instruction metering. New meter window displays crude total instructions executed and rough instructions per second.

• Built framework for varying display types (text, hires, super etc.)
• Added more VGC soft switches for video display mode handling.
• VGC was using location $0800 for screen storage. No effect, just wasted code. Fixed.
• Hires implemented. Now runs copy protected Alien Rain. Woohoo!

• Implemented ADB type ahead buffer (supports 256, but is equate change).
• Refactored breakpoints ready for masking support.

• Speed improvements for VGC. Screen is now about five times faster on simple character updates.
• Implemented full correct Apple IIgs key mappings in ADB, including modifiers, repeat and any key down. (Ugly in Unicode Java, but works well)

• 5.25" support integrated into IWM.
• IWM now supports everything except writing. (can read and move around on 5.25" and 3.5" nibblized images)
• VGC wasn't clearing the top line of the 40 column screen correctly. Fixed.
• New command line switch to load nibblized images into slots 5 and 6, drives 1 and 2.
• New "[modulename]s" command in xdebugger to show status of each module. So far only IWM supported.
• In a few circumstances, debugging access to I/O registers were actually triggering I/O. Fixed.

• Keyboard now supported through ADB. (Can now boot to DOS 3.3, write/run Applesoft code, drop to monitor and use GS monitor etc. First time in monitor looking at memory was spooky!)

• Added stepper motor support for 5.25". Now boots 5.25" nibblized disk images. (Which means there's now a whole bunch of software it will run)
• Now saves and restores BRAM to a file.

• Now reads/boots 5.25" nibblized disk images. (DOS 3.3 master boots until track seek required)

• Initial implementation of IWM with minimal support.

• Basic ADB functionality implemented. Now runs to slot 6 boot code and spins with no drive attached. Looking good.
• New command line parameter -logadb turns on ADB emulation diagnostics.
• Debugger now monos itself when doing a step over.
• Debugger now disassembles tool call names ala GSBug.

• Real time clock rewritten. Now supports all functions except for time (amusing if you don't know the Apple IIgs).
• New command line parameter -logclock turns on real time clock emulation diagnostics.
• Xdebug switch was turning internal debugging off instead of on. Fixed.
• When in run mode, debugger now "monos" itself to show that the various displays are no longer valid.
• Debugger wasn't able to read from $C010 to $C01F switches. Fixed.
• Debugger setting of breakpoint while CPU running wasn't updating the debugger display with the new breakpoint. Fixed.

• Can now set breakpoint counters instead of just defaulting to 0

• Changed debugger "Special" button to "Run". Breakpoints now provide the functionality of Special
• Application/main class name changed from JavaA2 to AppleIIgs

• Breakpoints now support GSBug style counters
• Register display is now larger and easier to read

• Moved development environment from CodeWarrior for Windows to BBEdit on Mac OS X
• Implemented MVN and MVP
• Added overflow support to ADC, SBC and BIT
• Can now set breakpoints and watchpoints from the debugger. Breakpoints now break, watchpoints don't yet watch
• Now displays version on debugger
• Version and code name no longer hard coded

• VGC fetch from text screen exception. Fixed

• Moved all preferences and startup arguments into preferences object
• Added -novbl argument to dummy vbl speed up
• Threads are now all named appropriately
• Debugger can now switch between various step/trace modes asynchronously
• Added initialisation complete message
• Cleaned up console displays
• BRK wasn't pushing status register in native mode. Fixed
• Implemented COP, RTI, STP and WDM instructions
• No addressing mode instructions now disassemble neater
• Added -logall argument to turn all logging options on
• Refactored logging management
• Re-enabled Debugger Trace now that Debugger is async (ie. it can be stopped while tracing)
• Refactored Supervisor, too many changes to mention. Supervisor = console debugging interface, similar to the Apple II monitor, but external to the emulator, so it is always available. Mainly a test bed for Debugger features before moving into the Debugger, but still available via -supervisor argument at start up
• Added external devo test support code

• Debugger stop was doing a step. Fixed
• Added -debugger, -help, -loginits, -log65816, -Xdebug, -nolog and -rom romname command line arguments
• Step out fixed. Two problems: Debugger display not updating, test case I was using never actually returns
• Special now updates Debugger displays after the trigger fires
• Operand fetch cache wasn't single touching. Fixed
• ORA (d,x) didn't have an addressing mode. Fixed
• ADC [d] wasn't registered as a valid instruction and didn't have an addressing mode
• XCE was shortening mx on switch to native. Weird this never showed up before. Fixed
• Debugger wasn't predicting mx switches for disassembly correctly. Fixed
• Debugger now shows start up message
• Absolute indirect addressing wasn't actually doing the direct part (COUT never calls COUT1 :-)
• JMP and JSR were only getting operand, not operand result. Fixed
• Absolute indexed indirect refactored
• Added ROM symbols (826 to be precise) for disassembly/Debugger (see screen shot for example)
• Removed wasted space in disassembly line
• Generally cleaned up Debugger display
• Skip wasn't updating instruction list with new location. Fixed
• First cut at VGC. Supports 40 column text with C022 text/background color support
• Implemented VBL with close to correct timings (standard Java won't do fractions of a millisecond, so is slightly longer than normal VBL. Could be refactored later to rectify using odd/even averaging)
• Emulator now exits when main Apple IIgs window is closed
• We now have a running Apple IIgs! Time for bed...

• Rewrote Debugger in Swing with custom JComponents. Improved speed of Debugger x10. (no shit!)

• JMP indirect was JMP absolute. Fixed
• Debugger special trigger was stepping an extra instruction. Fixed
• Now times execution blocks to nearest millisecond, although timings vary due to JVM having to deal with varying operating systems. For reset to $FC58 (on clock failure):
• Win98 = 5930ms (5.93s) -- Varies between 5880ms and 6100ms
• MacOSX = 410ms (0.41s) -- Initial execution (pre-HotSpot) = 599ms (0.59s)
• SBC was setting borrow around the wrong way. Fixed
• Added simple step over support to Debugger
• Adjusted some Debugger controls now that I've seen the layout on four platforms

• Debugger can now stop traces and step outs
• For some reason PER was stack relative. Fixed
• Initial implementation of dummy ADB GLU
• Now fully traces from power on until GS startup failure "sliding apple" screen

• Fixed bug where multi-byte operands were big endian
• Added "Skip" to Debugger
• Added trace to "Debugger/console/both/none" option
• Ten more instructions -- all main instructions complete
• All opcodes implemented
• All addressing modes implemented
• Fixed bug where most N flag setters were setting Z

• Some performance improvements in addressing modes and non-critical MMU use
• Now correctly touches 65816 address lines in read/write/read-modify-write modes
• Fixed console output prefixes to be more consistent
• Fixed "ROM not found message" to display correct ROM name

• Debugger now displays registers, stack, pre and post execution disassembly

• CPU engine now runs in its own dedicated thread
• Debugger now runs in its own dedicated thread

• Initial cut at optional AWT based interactive debugger. Required now that the CPU engine can trace deep into the ROM.
• Step / Trace / Trace out / Reset / Stop
• Optional step/trace disassembly

• 6 more addressing modes
• 1 more instruction
• Improved Debugger
• optional step disassembly
• step out support
• optional mmu monitor
• configurable break watch point

• 4 more instructions
• First implementation of clock
• MMU support for observer watchpoints (to implement I/O switches)
• Fixed direct page indexed wrap problem
• Debugger improvements

• 2 more addressing modes
• 9 more instructions
• Fixed offset problem in relative mode
• Fixed roll bug in ASL

• 5 more addressing modes

• 3 more addressing modes
• 16 more instructions
• Improvements to disassembler
• Bug fixes for two earlier addressing modes
• Bug fixes for stack misplacement problems

• Coding of initial CPU engine singleton
• 7 addressing modes
• 37 instructions
• stack implemented
• hardware reset and interrupt dispatching implemented
• all registers implemented
• Initial coding at MMU w/shadowing singleton
• Coding of interactive disassembler for CPU engine
• User selectable ROM

• Initial class design

Alien Rain on an iMac G5 2GHhz

Boot to DOS 3.3 (under Mac OS X 10.1.3)

Slot scan waiting for boot disk

Debugger (under OS X 10.1)

Failure intialising ADB