fix(rivetkit): treat empty actor input bytes as absent

[abcxff]

No description provided.

[abcxff]

Stack for rivet-dev/rivet

• #5539 feat(container-runner): report actors as crashed on unexpect …
• #5536 feat(container-runner): log unexpected platform SIGTERM as a …
• #5530 fix(rivetkit-core): cancel driver alarm before sqlite teardo …
• #5531 feat(container-runner): keep instance warm instead of self-e …
• #5533 chore(container-runner): default stop grace to 10s
• #5528 fix(rivetkit): log and surface actor startup failures
• #5529 fix(rivetkit): treat empty actor input bytes as absent 👈
• main

Get stack: forklift get 5529
Push local edits: forklift submit
Merge when ready: forklift merge 5529

_{change ouvoqtns}

[railway-app]

🚅 Deployed to the rivet-pr-5529 environment in rivet-frontend

Service	Status	Web	Updated (UTC)
frontend-cloud	😴 Sleeping (View Logs)	Web	Jul 31, 2026 at 4:22 am
frontend-inspector	😴 Sleeping (View Logs)	Web	Jul 31, 2026 at 4:21 am
kitchen-sink	😴 Sleeping (View Logs)	Web	Jul 31, 2026 at 4:19 am
website	😴 Sleeping (View Logs)	Web	Jul 31, 2026 at 4:18 am
ladle	✅ Success (View Logs)	Web	Jul 31, 2026 at 4:13 am
mcp-hub	✅ Success (View Logs)	Web	Jul 31, 2026 at 4:13 am

[claude]

Review

Small, well-scoped fix that brings Input<A>::decode()/decode_or() in line with Snapshot::decode() (which already filters empty bytes at start.rs:108) and decode_conn_params (which already special-cases empty bytes at start.rs:529). Confirmed the TS side (native.ts's decodeValue) already treats empty-but-present input bytes as absent, so this change closes a genuine Rust/TS parity gap rather than introducing new behavior.

Potential issue

• is_present() (start.rs:44) still checks only self.bytes.is_some() and was not updated alongside present_bytes(). After this change, a caller can see is_present() == true for a zero-length-but-Some payload, then call decode() and get Err(MissingInput) (for non-unit input types) or the unit/default value instead of an actual decoded value. "Present" no longer implies "decodable as real data." Since Input is public API in the (preview) rivetkit Rust crate, external callers could reasonably guard on is_present() before calling decode(). Worth considering whether is_present() should also filter empty buffers (delegate to present_bytes().is_some()) for consistency, or document the distinction if it's intentional.

Nits

• decode_conn_params (start.rs:528) also treats a CBOR-encoded null ([0xf6]) as "use default," but Input::decode/decode_or do not extend the same treatment to a CBOR null payload for actor input. Likely out of scope for this fix (the PR is specifically about the empty-buffer case), but flagging in case the engine can ever send a CBOR null for input the same way it can send an empty buffer.

Test coverage

Good. New tests cover both the "empty bytes defaults" path (input_decode_or_default_treats_empty_bytes_as_missing) and "empty bytes maps to unit" path (input_decode_treats_empty_unit_as_unit), plus a regression test ensuring non-empty invalid CBOR still hard-fails (run_actor_invalid_input_fails_to_start) so the fix doesn't accidentally widen into silently swallowing bad input.

Style

Comments follow the repo convention (complete sentences, no dashes/em-dashes) and the fix mirrors the existing filter idiom already used by Snapshot::decode. No concerns there.

Overall this looks correct and safe to merge; the is_present() inconsistency is the only thing I'd want a second look at before landing.