GCBASIC Code

Restore the cache value correctly in HPWMUpdate

Compiler additions:
- Fixed bug where temporary string variables were overwritten prematurely
- Added asm ShowDebug command - anything following this will have any constants replaced, and will then be inserted as a comment into the assembly at that location
- Replaced debugging nops in system.h with asm ShowDebug
- Temporary string variable sizes are now set based on the string to be copied into the variable - a temporary variable to store "Hello, world" would now be 13 bytes rather than 20 or 40 (depending on the chip).
- Command line /K:A switch now correctly passes all input lines to assembly file. Comment lines were omitted erroneously ever since this switch was implemented.
- Fixed bug in Rotate command which caused declarations for non-byte variables to be removed from the assembly when only the Rotate command operated on that variable

PWM16 for the Microchip 16bit PMW modules!
updated lowlevel.dat
See https://github.com/Anobium/Great-Cow-BASIC-Demonstration-Sources/tree/master/PWM%20Solutions/PIC%20PWM16%20Examples for code

Tidy up by deleting all the demos - they are now secure in GitHub

Updated.dat files with new chip data and K42 support

Update to lowlevel libraries to support 18F2xK42 chips
A multiple upload but these are all relative to the K42 chip set.

More additions to compiler:
- Addition and subtraction now implemented for double and single precision floating point variables. Other arithmetic still unsupported.
- Partially added ULongInt (unsigned 64 bit integer) and LongInt (signed 64 bit integer) types. For these, addition/subtraction, rotating, boolean operations, and bit setting and testing are currently supported. Other operations are missing and will be implemented as needed or after floating point support is complete.

More compiler changes for floating point (still not usable):
- Added conversion routines to convert between integer, single and double variables. All required conversion routines should now be present.
- Compiler now looks for subroutines to do floating point addition, subtraction, multiplication and division, but these are not yet implemented
- Rewrote code in compiler that handles Repeat command, functionality is the same but new code is much more compact and can be expanded to handle more bytes easily.

Further work on floating point, still not usable. Additions:
- Conversion between any combination of byte, word, long, single and double now possible
- Rotate command now accepts casts

Floating point support work in progress:
- Can now create Single and Double variables
- Can set these variables to fixed values
- Can copy between variables of the same type (single to single, double to double)
Checking in for safekeeping, much more work needed before any of this is useful. Compiler is still safe to use as before, but do not attempt to use floating point.