Revises the v1 mint rule after the maintainer catch that its key scoped
families to the profile bundle (printer brand), fragmenting one commercial
product into N ids (PolyLite PLA: five bundles, all filament_vendor
"Polymaker", up to five ids).
Architecture: BBL and QD_* stay frozen islands; OrcaFilamentLibrary becomes
the single declaration point for every other material family; vendor
bundles carry only same-alias specializations (no filament_id key) that
shadow the OFL preset per printer and resolve its id through the loader
walk. New mint key: uuid5 over
"filament_product/<filament_vendor>/<filament_type>/<family_name>" from the
root's flattened config — bundle-independent (hoisting families into OFL is
id-stable), and the type component keeps the four known same-name-
different-type groups apart until their data bugs are fixed.
Content-addressing replaces "ids immutable once shipped": identity edits
re-id the family, made safe by turning retired_filament_ids.json into a
shipped succession ledger (old id -> successor, chains allowed,
cross-island hints permitted) consulted on resolution miss in the AMS sync
and lookup paths. The MoonrakerPrinterAgent hardcoded generic-id map is
replaced by a runtime preset lookup, removing the code/profile lockstep.
Migration: v3.0 tooling + W1 client hardening + W3 type fixes; v3.1 OFL
re-mint with succession entries; v3.2 vendor bundles (391 unshipped OF ids
re-derive without retirement; generic tunings re-point to OFL; ~800 shipped
legacy ids re-mint with succession, incl. the 57 multi-vendor GF residue
and the 10 P-hex system ids, which also exits them from the destructive
check_ams_filament_valid P-gate); v3.3 optional id-stable consolidation
into OFL. Gate battery unchanged plus succession-resolution tests.
filament_id_plan_v2.md gets a status note: its P+md5 verdict stands; its
work items are absorbed or superseded by v3.
Examined minting system filament_ids with CreatePresetsDialog.cpp:487's
user-custom allocator (adopt-by-base-name, else "P"+md5(name)[0:7]) for
Bambu AMS-sync compatibility. Verdict: keep the OF* mint; capture the
proposal's value through bounded work items instead of a re-mint.
- The device path never validates id shape: tray_info_idx is an opaque
string end-to-end (DeviceManager.cpp:1642; DevFilaSystem.cpp:512-514)
and firmware persists arbitrary bytes (bambulab/BambuStudio#5436).
Resolution is by value against the GF catalog plus the account's cloud
custom cache; unknown ids show "?" regardless of shape, and system
presets can never enter that cache (Preset.cpp:2071 gates upload on
is_user()).
- The only id-shape dispatch in the tree is destructive for P-shaped ids:
check_ams_filament_valid (DeviceManager.cpp:5335/5352/5395/5410) remotely
clears an AMS tray (:5345) or rewrites its temps when a P-shaped tray id
drops out of the user-root preset list — a state name-keyed minting
creates by construction. GF* and OF* ids are structurally immune.
- The real value is captured without re-minting 391 unshipped ids across
1327 files: the adopt step == curated GF adoption via the snapshot ledger
(W4); the hash belongs to user presets (W5, cf. PR #13315); client
hardening (W1) that the ten already-shipped P-hex system ids (Cubicon x8,
Ginger, Artillery) need today anyway.
Method: 6 parallel evidence agents (generator semantics, AMS/device path,
exhaustive shape-dispatch sweep, upstream + online sources, tree-wide
dry-run over 5892 presets / 1148 base names, branch change inventory) into
a 3-judge panel (keep-OF* won 23/40 aggregate); the upstream generator was
verified byte-identical to BambuStudio master, and every load-bearing
file:line and number in the document was re-verified against the tree
before commit.
