* Introduce assets/_extracted/VERSION, with text extracted there
* move to `extracted/text/`
* Update gitignore s
* rework args for msgenc.py
* put mkdir with others, until theyre all moved at once
* move 0xFFFC back to being extracted, making it use specific macro `DEFINE_MESSAGE_NES` to handle its special behavior
* prettier gitignore
* Move messages 0xFFFC, 0xFFFD to committed message_data.h
Currently the C preprocessor is used to remove comments from the text
files before encoding them.
However the option used (`-fpreprocessed`) is only supported by gcc –
but not by clang, which breaks the macOS build.
By using Python to remove comments, we can remove the use of this
gcc-specific option, and fix the macOS build.
* Initial progress on z_message_PAL, very messy
* Fix merge
* Some more progress
* Fix merge
* More z_message_PAL
* Small progress
* More small progress
* message_data_static files OK
* Prepare z_message_tables
* Matched another function, small updates
* Attempt to use asm-processor static-symbols branch
* Refactor text id declarations
* Begin large text codes parser function
* Fix merge
* Refactor done
* Build OK, add color and highscore names
* Remove encoded text headers and automatically encode during build
* Fix kanfont
* Various cleanups
* DISP macros
* Another match aside data
* Further progress
* Small improvements
* Deduplicate magic values for text control codes, small improvements
* Tiny progress
* Minor cleanups
* Clean up z_message_PAL comment
* Progress on large functions
* Further progress on large functions
* Changes to mkldscript to link .data in the .rodata section
* data OK
* Few improvements
* Use gDPLoadTextureBlock macros where appropriate
* rm z_message_tables, progress on large functions
* 2 more matches
* Improvements
* Small progress
* More progress on big function
* progress
* match func_80107980
* match Message_Update
* match func_8010BED8
* done
* Progress on remaining large functions
* Small progress on largest function
* Another match, extract text and move to assets, improve text build system
* Small nonmatchings improvements
* docs wip
* Largest function maybe equivalent
* Fix merge
* Document do_action values, largest function is almost instruction-matching
* Rename NAVI do_action to NONE, as that appears to be how that value is used in practice
* Fix merge
* one match
* Last function is instruction-matching
* Fix
* Improvements thanks to engineer124
* Stack matched thanks to petrie911, now just a/v/low t regalloc issues, some cleanup
* More variables labeled, use text state enum everywhere
* More labels and names
* Fix
* Actor_IsTalking -> Actor_TalkRequested
* Match func_8010C39C and remove unused asm
* More docs
* Mostly ocarina related docs
* All msgModes named
* Fix assetclean
* Cleanup
* Extraction fixes and headers
* Suggestions
* Review suggestions
* Change text extraction again, only extract if the headers do not already exist
* Fix
* Use ast for charmap, fix assetclean for real this time
* Review suggestions
* BGM ids and ran formatter
* Review comments
* rename include_readonly to include_data_with_rodata
* Remove leading 0s in number directives
* Review suggestions for message_data_static
* textbox pos enum comments, rename several enum names from Message to TextBox
Co-authored-by: Thar0 <maximilianc64@gmail.com>
Co-authored-by: Zelllll <56516451+Zelllll@users.noreply.github.com>
Co-authored-by: petrie911 <pmontag@DESKTOP-LG8A167.localdomain>
Co-authored-by: Roman971 <romanlasnier@hotmail.com>
