Apply the same lazy-factory pattern to FSLaunchboxHandler and FSSyncHandler
that ssh_sync_handler now uses. With both opt-in features deferred to
first-use, the tolerate_missing_base escape hatch on FSHandler is no longer
needed — every handler now fails loudly on mkdir failure, which is the
right behavior for the always-on core paths (assets, library, resources).
Touched call sites:
- resources_handler._resolve_local_file_uri (launchbox)
- sync_watcher.py, endpoints/device.py, tasks/manual/sync_folder_scan.py
(fs_sync)
Net effect:
- Default installs never poke /romm/launchbox or /romm/sync at startup.
- Misconfigured opt-in users get a clear, actionable PermissionError at
the call site instead of a silent warning followed by mystery failures.
- tolerate_missing_base, its tests, and one stale log import are gone.
Importer (gamelist/launchbox file:// flows) and exporters (gamelist.xml,
metadata.pegasus.txt local exports) now hardlink media assets when source
and destination share a filesystem, falling back transparently to a copy
on EXDEV / EPERM / EOPNOTSUPP / EMLINK / EACCES (cross-device, FAT32,
exFAT, network mounts, etc.). Saves disk space and is effectively
instantaneous on large files (videos, manuals, miximages).
Covers keep a real copy (allow_link=False) because _store_cover resizes
the small cover in place via PIL.Image.save, which would truncate the
shared inode and corrupt the user's source image.
Also makes FSSyncHandler tolerate a missing/unwritable /romm/sync at
startup: an OSError from mkdir now logs a warning instead of crashing
the whole app at module-import time. Sync calls still fail at use time
if the mount remains broken — the right place to surface the error.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>