continuum-proxy and continuum-backend each had their own idea of what the
edge WebSocket carried, and the two did not overlap anywhere: the backend
expected an `auth` handshake the gateway never sends, telemetry shaped one
printer per message where the gateway sends a batch, and UUID printer ids a
gateway has no way to know. Neither end had a command path that matched the
other. Nothing failed, because nothing was written down for both sides to
mirror. This is that document.
schemas/continuum/v1/edge/ is canonical: client_message, server_message,
printer_telemetry and printer_command. packages/ts-types gets its first
real TypeBox mirror - the placeholder there said to add one "once something
actually consumes it", and continuum-backend's edge route now does. The
Rust mirror stays in continuum-proxy/src/uplink/protocol.rs rather than
packages/rust-types, since the gateway is its only Rust consumer.
Two decisions worth recording:
Field names are snake_case, unlike the backend's HTTP routes. The other end
is serde-derived Rust, and snake_case is what those structs produce without
a rename attribute on every field. It is also what the rest of this
repository already uses.
Optional fields accept an explicit null as well as being absent. A gateway
omits what it has no value for, which is what these schemas describe - but
serde emits null by default unless every field is annotated, and a single
such field would otherwise fail validation and take a whole farm's
telemetry batch down with it. Strict in what we send, liberal in what we
accept; the contract tests on both sides pin down each half.
Also bumps this package's @sinclair/typebox to ^0.33.12 to match what
continuum-backend already pins. It was on ^0.32.0, which resolved to a
separate 0.32.37 install - two TypeBox versions means two incompatible
TSchema identities, and Elysia rejected every schema from here on those
grounds alone.
Continuum is intended to be sold, so every dependency has to permit
distribution in closed-source commercial software. Nothing recorded that
before: our own crates and npm packages carried no licence field at all,
which made `cargo deny` report them as `unlicensed` errors and left the
actual policy nowhere.
Crates now declare `LicenseRef-Proprietary` with `publish = false`.
That string is deliberately not a real SPDX identifier, so it must never
appear in a `deny.toml` allow list - a third-party crate claiming it
would be a genuine finding. npm packages use `"license": "UNLICENSED"`,
the equivalent convention there.
`deny.toml` holds the permissive-only allow list. The policy itself, the
audit, and the three categories cargo-deny cannot see (external
binaries, ML model weights, self-hosted servers) are documented in
continuum-common's docs/LICENSES.md.
They described the old flat continuum/v1 shape, replaced by the
material/printer/vendor restructure. Nothing outside this repo imported
them (checked continuum-proxy, continuum-ai-worker, continuum-backend,
continuum-app). Left both packages in place as empty shells - add a
hand-written mirror back per new schema file once something actually
consumes it.
Nothing imported the proto-generated code (zero references to
continuum-rust-types or @continuum/ts-types's proto export in any
consumer repo), and protobuf's real value - one schema compiled to a
compact binary wire format for both languages - never applied: BE<->proxy
and FE<->BE are both JSON. Also fixes 8 broken $ref paths in the bambulab
schemas and a root tsconfig.json that was sweeping the gitignored
examples/ directory into typecheck.
- proto/continuum/v1/*.proto -> schemas/continuum/v1/*.schema.json
- packages/rust-types: hand-written serde structs instead of
prost-build codegen
- packages/ts-types: drop the generated/ proto barrel export, keep the
hand-written TypeBox schemas as the TS source of truth
- root package.json: drop build:ts, protoc, ts-proto
- v2/common/device/device.schema.json: stray backtick in bed $ref broke resolution
- v2/report/print/print.schema.json: ams $ref pointed at wrong path (missing ams/ subdir)
- v2/common/device/plate.schema.json: $id was copy-pasted from bed.schema.json
- Add v2/request.schema.json and v2/report.schema.json dispatchers - v2's leaf
schemas were unreachable without them, same oneOf convention as v1
- Rewrite schemas/bambulab/README.md layout section, which still described
v1's pre-rename paths and claimed v2 was removed
- PrintJob.status serialized as a raw i32 in Rust (prost stores proto3 enums
as i32) while ts-types' TypeBox schema validates full enum name strings -
add serde with= on the field so Rust JSON matches TS and protobuf's own
canonical JSON enum mapping, plus a regression test
- print.schema.json, _ams.schema.json, _ams_unit.schema.json still
pointed at the pre-rename filenames (ams.schema.json, ams_unit.schema.json,
ams_tray.schema.json) after the underscore-prefix split - fixed all
three, verified with a full jsonschema + referencing validation run
against real captures (get_version, pushall, a full print/ams/xcam
report with both empty and loaded trays) plus negative tests.
- _upgrade_state.schema.json: fixed its $id (pointed at report/info/,
file lives under report/print/) and its title (said "ModuleInfo",
clearly copy-pasted from _module_info.schema.json). Wired it into
print.schema.json as the print report's `upgrade_state` property -
the file existed but nothing referenced it yet.
- Deleted schemas/bambulab/v2/, which had been recreated as stale,
broken copies of an older v1 (missing the envelope allOf, missing
the underscore renames, $ids still pointing at v1/ paths) - the
repo's own README documents v2 as intentionally removed until V2's
wire format is known to differ from V1's.
- Updated schemas/bambulab/README.md's layout tree and inheritance-
example reference, both stale after the underscore rename.
- Removed stray tsc -b build output (packages/ts-types/src/*.d.ts/js)
that isn't meant to be committed (outDir is dist/, gitignored).
Verified: cargo check clean (continuum-proxy, continuum-ai-worker,
rust-types), tsc clean (continuum-backend, continuum-common ts-core),
10/10 JSON Schema validation cases passing.
Reconsidered after actually feeling the friction: this data is JSON over
MQTT (not binary protobuf), only continuum-proxy ever parses it (raw
vendor wire format, never crosses into TS), so none of protobuf's actual
value (cross-language codegen, binary wire efficiency) applied here.
Every gotcha hit while building this (field_attribute's oneof/variant
path collision, needing .bambulab-scoped type_attribute calls, explicit
#[serde(default)] per not-always-present field) came from forcing
prost-build's codegen model onto a plain-JSON, Rust-only use case.
Verified the replacement is strictly simpler, not just smaller: a
hand-written struct's Option<T> field needs zero extra attributes to
handle a missing JSON key (prost's generated String needed an explicit
field_attribute call per field), and #[serde(flatten)] on a hand-written
field has no ambiguity at all (prost-build's oneof/variant path collision
doesn't exist without prost-build's codegen in the picture).
The replacement lives in continuum-proxy: src/printer/bambu_commands/.
continuum.v1 is unaffected — that schema is genuinely cross-language and
proto still earns its keep there.
The rest of proto/bambulab/ was pure boilerplate/comments with nothing
concrete to copy from. This one message (sequence_id + command, matching
Bambu's real get_version request) is wired all the way through: real
proto, real generated Rust struct, lib.rs actually include!'d instead of
commented out. continuum-proxy's bambu_protocol_demo example serializes it
and diffs the output against Bambu's actual wire JSON byte for byte.
Verified with cargo build (this crate) and cargo run --example
bambu_protocol_demo (continuum-proxy) - output matches exactly.
Adds bambulab/v1 and bambulab/v2 to the prost-build pipeline, scoped
separately from continuum.v1's camelCase convention: bambulab.* uses
snake_case, matching Bambu's actual MQTT wire JSON exactly, since this is
edge-only traffic (continuum-proxy) that has to match the printer's real
format rather than a schema we get to design.
Verified the scoping doesn't collide: type_attribute calls are cumulative,
so the existing '.' blanket selector had to become '.continuum' specifically
- keeping it global would stack a second, conflicting #[serde(rename_all)]
onto every bambulab type too and fail to compile.
No message bodies added - proto/bambulab/{common,v1/*,v2/*}.proto are
intentionally just syntax+package+guidance comments (matching what was
already started for v1). build.rs has inline notes on the two real gotchas
already hit while prototyping this: the field_attribute path-matching
oneof/variant collision (solution: hand-write the envelope enum in Rust
instead of modeling it as a proto oneof), and serde needing #[serde(default)]
on fields that aren't always present.
protoc warned that print_job.proto imported template.proto without using it
- removed. ts-types/package.json was missing protobufjs and long, which the
ts-proto-generated code imports at runtime - tsc failed with 'cannot find
module'. src/index.ts also re-exports './generated' as a namespace, but
ts-proto wasn't configured to emit an index barrel file for it - added
outputIndex=true to the protoc invocation.
Verified with a clean 'bun run build' (both TS and Rust codegen) and
'tsc -b'.
Protobuf data model (Template/PrintJob/MaterialProfile/PrintParameters)
with codegen targets for TypeBox+TS types (packages/ts-types) and
prost-derived Rust structs (packages/rust-types).
octoturge