The review of the commit that taught nx2navi.ini's readers off, false and no found three readers it left behind. cfg.c
lowend_of reads low_end_gpu through raw_get, which skips switch_word, and still took anything but "0" as on: with
"low_end_gpu = off" cfg_get, cfg_bool and the menu said off while cfg_low_end_active() was 1, so the mode's defaults
and its LOWEND_ONLY keys stayed in effect, and F-7's check, which now accepts the word, logged nothing. lowend_of now
takes the value through switch_of as cfg_bool does: 0 and the off words are off, anything else set is on (an
environment value "0 ; x" stays on, as cfg_bool reads it). tools/make_game.py read the file on its own too:
shader_carry took only "0" as off and kept a comment in the value, so with "shader_carry = off" or "shader_carry = 0 ;
note" an update skipped the full shader preparation and left the shaders to a game that no longer carries them over;
guest_opt_ini took low_end_gpu as on for anything but "" and "0". shader_carry now reads the switch as the game does
(the environment wins over the file and is taken as written, the file's last line with a value wins, a comment after
a blank is cut unless the value is in double quotes, make_game.py ini_choice, and 0, off, false and no in any case are
off); guest_opt_ini keeps its cut and takes the same words as off.
tests/cfg_parse_check gains a lowend case (low_end_gpu = Off, = 0 ; note and = yes from the file, FALSE and "0 ; x"
from the environment; two of them fail on the old lowend_of), and launcher_test's MakeGameTest two tests of
shader_carry's and guest_opt_ini's reading, which fail on the old make_game.py. cfg_rows_check reports 0 problems, and
the -Wall compile of cfg.c shows no warning.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>