- added universal symbols file
- chip memory definitions no longer include symbol definitions.
they use the new universal symbols file
o cpu
- moved instruction_result from cpu package to result package
o disassembly
- style parameters moved to new result package
o memory / debugger
- improved PEEK command implementation
- can now use symbolic addresses
- added ONSTEP command. similar to ONHALT but run every step
- implemented LAST command. debugger no longer runs the result
of the last instruction unless it is part of the ONSTEP
command sequence
- fixed printing of HEX. step mode video no longer crashes if
print style includes hex.
o video
- sketched in collisions support
o cpu
- BIT instruction fixed (zero flag is result of AND with
accumulator but sign and overflow flags are set with
reference to the value from memory)
- implemented video in part
- implemented sprite framework - ball sprite completed
- reorganised existing tia code in repsonse to new code
o debugger
- tweaked existing commands and how they work
- filled in more of the TIA implementation
- RSYNC works correctly
- polycounter and other components altered as required
o television
- sketched in television interface
- implemented dummy television
o debugger
- syntax normalisation -- allowing variations in how we specify hex numbers
- added VERBOSE command
o renamed mflib package to assert
- Assert function renamed to CheckValueVCS
- added MEMMAP command
o tidy up
- changed pointer recievers to value receivers when the method does
not mutate the type instance
- I'm not sure about this but I like if for now - this way it is
clearer, to me at least, which methods mutate and which don't
o memory
- small reorganisation of memory package in preparation for
implementation of the video chips
o renamed project to "gopher2600"
- replaced both register implementations with just one that uses
32 bit integers as the underlying implementation. very smart and
very quick by comparison to the bit array implementation.
- removed go routine implementation of cycle stepping; far too slow
- replaced with a callback implementation. I'm thinking that it might
not be as flexible but that in the long run that it doesn't matter.
besides, the speed difference is definitely worth it.
- reorganised register implementation
- existing implementation now call rbits
- added an implementation called r16bits, using uint16 intead of a
bit array
- only implemented Load(), Add() and the To*() methods
- just enough to speed up programme counter operations
- now sends a reference to InstructionResult over the result channel,
rather than a copy
o cpu
- adjusted register declarations to match changes in registers
implementation
o mflib
- consolidated assert() function into one package
o debugger
- sketched in debugger
- we now start one goroutine at cpu intialisaton and not one for each
instruction
- caching of all 8 bit and 16 bit patterns so we don't have to
generate them every time
- improved FPS between 70% to 100%
- added basic input loop
- primitive but effective FPS test
- added cpu profiler generator
o vcs
- cartridge attachment working properly
- now read reset address when machine is reset
o cpu
- fixed branching
- backward jumps weren't working
- cycle count sometimes wrong
- fixed cycle counts for pre-index and post-index addressing
o
- refinements
- added white-box testing
o cpu
- moved cpu definitions into its own package
- no longer requires csv file at run time
- csv parser moved into a code generator, embedding the data into
the code
- needs refinement
- StepCycle to be be used if cycle level of granularity is required
- StepInstruction provided if coarser control is preferred
- definitions.csv corrected
- behaviour of status register ToUint8() corrected
o memory
- MockMem now inherits from main memory implementation and adds
functionality where required
- 6502 implementation is basically correct but does not handle
- interrupts
- pre-fetch cycle
- anything less fine grained than instruction stepping
- memory sub-system sketched in
o retroactive adding of GPL copyright notice
