v0.8.0

Happy Pride month! 🏳️‍🌈

This release internally refactored a lot of C code into more idiomatic C++ (partly to fix memory leaks), which *hopefully, knock on wood, fingers crossed* won't have broken anything else.

We're optimistic that the next release will be a 1.0.0 rewritten in Rust. 🦀

Added:

• RGBGFX allows #none as a "gap color" in -c palettes (thanks, @eievui5!)
• RGBFIX -L/--logo can specify a custom logo 1bpp instead of the Nintendo logo
• RGBASM allows "\0" as a NUL character in strings
• RGBASM align 16, N acts like org N (so align 16 acts like org 0)
• RGBLINK -p padding applies when output goes past the -O overlay file contents

Removed:

• Defining symbols without using DEF
• RGBASM -H/--nop-after-halt, -l/-auto-ldh, -h/--halt-without-nop, and -L/--preserve-ld options (the auto-ld-to-ldh and auto-nop-after-halt behaviors are now completely removed)
• RGBASM -i for --include (use -I)
• RGBGFX --output-* long options (use --auto-* or the short -A/-P/-Q/-T options)

Changed:

• Strings and symbol names are no longer limited to 255 characters!
• expr && 0 and expr & 0 are constant 0; expr || 1 is constant 1
• All warnings/errors refer to input/output file "-" as "<stdin>/<stdout>"
• Documentation switched from reStructuredText to Markdown for ease of contributing

Fixed:

• Increment object file revision number to 10 (since 0.7.0 updated the object file format but didn't increment this)
• Linker scripts infinitely looped when parsing decimal numbers (thanks, @huderlem!)
• Long RGBASM interpolation format specs could segfault
• RGBGFX --reverse --columns was incorrect for non-square images (thanks, @sfiera!)!!!!
• ACO palette files were not being read correctly
• Specifying an embedded palette to override a previously specified palette crashed (thanks, @ZoomTen!)
• Warning/error file-stack traces for macros inside REPT/FOR loops are complete
• Building RGBDS nested in another Git repo won't pick up that repo's tags for the RGBDS version
• Some fixed-point functions could give inconsistent results if they internally gave infinity or NaN (e.g. FMOD(value, 0.0))
• Formatting $8000_0000 (-2,147,483,648) as a string was omitting the negative sign
• BANK(NotALabel) is 1, not $4B4E4142 (the numeric value of "BANK" 😂)
• \@ could change within a macro or REPT/FOR if it was first used within an INCLUDE
• Some built-in symbols could be misreported as being defined on the command-line