* Added a directive to control merging of SECTIONs: MERGE
* Added an indexed TEXT format that uses a String Symbol for the
indexing order
* General coding style consitencies
Pool are extended to:
* A pool label can be any size by using POOL <label>.<size> where <size>
is a decimal number
* A pool can define a pool within itself <pool> pool <name> <size> where
<pool> is the name of a pool that would fit another pool and <name> is
the name of the new pool and <size> is the size of the new pool
* pool labels are local if they are preceeded by . or ! or $ or @
* pool labels can be global
* label pools don't go out of scope so they can be declared in an
include file and used by the includer file
* label pool labels can be either global or local
* code fixes
* updated struse
* updated binaries
- Added a scope increment when entering a block of repeated code so only
the local symbols within the block would be destroyed each iteration
- Switched over to inttypes.h to make code a smidgeon more readable
- Flushing local labels after each rept/LUP source segment so local
labels within won't cross over the segment boundary
- The command line argument -org will force fixed address rather than
suggest a location, which in turn makes the -lst command line more
readable.
- Structs can contain empty lines
- Structs can be empty
- All structs have a size member named bytes, the number of bytes used
by a struct is <struct>.bytes
- Adding standard macros with for loops, memory copy, add, subtract,
move and set
- macros can be named with dots
- double negatives won't cause errors in expressions
- vice output will convert labels named "debugbreak" to vice breakpoints
rather than vice labels
- fixed issues with mixing conditional operators with math operators in
expressions
- struse.h replace with bookend didn't check bookends in all cases
- macros can be inside of conditionals (within if/else/endif, etc)
- string symbols mistake caused garbage code (missing braces around
assignment and return)
- rtl instruction had wrong opcode
- disassembler works a lot better with correct rtl, even for 6502
- disassembler can specify n bytes of data before code starts in src
mode (data=initial data size)
- String Symbols can be evaluated as expressions or assembled as code
- String Symbols can be generated by macros
- Cleaning up first page
- Adding more to the x65.txt documentation
- Started text documentation of assembler
- Built binary zip files (Windows)
- Fix for conditional + scope characters
- Fix for empty section "default" defaulting to BSS
- Using a section named DirectPage_Stack to adjust same named segment in
OMF file
- Resolving all zero page sections to fixed addresses before exporting
OMF
- Disassembler does more shenanigans to output more assemblable code
- Automatically determine the code address for c64 prg files
- Code and data segments are not just merged together, each segment in
each export group just gets address assigned and relocs / labels are
fixed up.
- BSS sections won't generate binary data if not inbetween code and data
segments for an export group (labels and relocs are still extended)
- DC directive now allows for .t (3 byte values) and .l (4 byte values)
decoration
- .lst includes all section info and address ranges if fixed
- Updated x65 dump tool with fixes to x65 data
- REPT caused a crash if expression was resolved after assembly
completed
- Fix expression % as a label at end of scope if another scope ended
before the current scope
- Removed # from conditionals since that doesn't fit with any assembler
style
- Added .l (long) and .t (triple) to label pool sizes, also supporting
.d as double which is the same size but different meaning than .w (word)
- XREF blocked linking with same name label
- REPT also works as a symbols for repeat counter to reduce need for
extra counting labels.
- "<<" got interpreted as a right shift
- Added LONG keyword to declare 32 bit values
- Each scope in a file shows the cycle count added up for instructions
within it.
- Added Merlin version of XREF which is EXT (was previously ignored)
- Added Merlin cycle counter on/off directive (CYC)
- XREF directive now required for referencing XDEF symbols from other
files
- Created cycle timing table for 6502, displaying for each line in .lst
files
- Map symbols might not have been updated with their labels
- Relocation index may be off
- Map symbols stored in 16 bits causing truncation of value in symbol
output
- Added a tool to dump the values in object files
- Fixed issue with CPU indexing
- Added missing 65816 jmp (addr,x) and jmp $123456,x and jsl
- Added a method for Merlin style linking to work in x65
- Made low byte / high byte (<, >) very low precedence
- Some more Merlin directives
- Split out some directive code from the huge ApplyDirective function
- Add source lines of keywords and labels to the output listing file
- x65 object file fixes
- added -acc=16 and -xy=16 command line options for 65816 immediate mode
instruction size
- added conditional expressions (evaluates to 0 for false and 1 for
true)
- fixed macro substitution to be whole word only
- IMPORT directive added, works like Kick Assembler
- -endm option also applies to REPT
- REPEAT directive is an alias of REPT
- Merlin macro fix uses ]1, ]2, .. as macro params
- Erroneously flushed local labels when starting to assemble a macro
- Added aliases to opcodes
- Fixed XCode warnings
- Revisited outdated comments
- Fixed CPU / PROCESSOR keyword giving an error even if the argument was
valid
- Rearranged switch statements to be in a more logical order
- don't allow ! in merlin labels
- don't skip initial : in merlin labels
- symbol files resolves labels from incobj relative sections
- assembler listing skipping disassembly for lines that was not built
from source code
- using -lst the disassembled and assembled source is written out side
by side
- option -opcodes saves out a source file with all opcodes & addressing
modes for testing
- cleaned up opcode table to be simpler and more robust so I can
consider supporting other cpus
- bug fixes
