mirror of
https://github.com/Klipper4FlashForge/firmware.git
synced 2026-10-23 13:52:16 -05:00
Oleksandr ShyshatskyiandClaude Opus 5
9b1f944293
The package installed every file the mod needs and then relied on the user
to hand-edit printer.cfg with seven [include] lines. That was the one
install step nobody could automate, because printer.cfg is the user's file
and run-append.sh may not write it -- so a freshly flashed printer had all
seven configs sitting in /usr/data/config/ and not one of them loaded, and
came up as a plain printer with no toolchanger and nothing saying so.
printer.base.cfg is ours, and the stock run.sh force-copies it to
/usr/data/config/ on every flash -- the same directory as the ff-*.cfg, so
relative includes resolve. Putting them at its end wires the mod up at
flash time and makes the undo button complete: flashing the stock
FlashForge package restores its own printer.base.cfg and the includes go
with it, leaving the ff-*.cfg as inert files rather than as a config
referencing extras that are no longer in /usr/prog.
It also retires the trap in the old instructions. "Append to printer.cfg"
is wrong for any printer that has run SAVE_CONFIG: an [include] below the
"#*#" marker makes Klipper read the whole autosave block as corruption
("modifications after header") and discard every calibrated value -- PID,
mesh, shaper, nozzle and dock positions.
Most of the documented ordering rules turn out not to exist. Nothing needs
to follow [virtual_sdcard]: virtual_sdcard registers SDCARD_PRINT_FILE and
M23 in __init__, ff_print takes them over at klippy:connect, and by then
every section is loaded. That was true of the old gcode_macro wrapper and
stayed in the docs after the move to ff_print.py. gcode_macro's
rename_existing defers to klippy:connect too, so the G28 and
SHAPER_CALIBRATE wrappers do not need their targets declared first either,
and ff-print-macros.cfg reads _FF_FILAMENT through a runtime template
lookup rather than at parse time. What survives is override order:
ff-runout.cfg after printer.filament.cfg's sensors, ff-chamber.cfg after
printer.chamber.cfg. End-of-file satisfies both.
ff_print.py's error message said "must be loaded after [virtual_sdcard]",
which pointed at the wrong problem; it can only fire when there is no
[virtual_sdcard] at all.
Both halves are checked on the replica now: case-install.sh asserts the
includes and files are present after a flash, case-recovery.sh asserts a
stock flash takes the includes away again. test_includes.py covers the
wiring in the repo -- every shipped ff-*.cfg included, every include
resolving, and the block after everything it overrides.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>