Context
Continuum owns shared boundary language and proof posture across the stack. It should not own runtime internals, storage, or compiler implementation, but its authored shared families need a deterministic representation profile so generated artifacts can travel without drift.
The active split is now:
- Continuum owns shared protocol nouns and authored family schemas.
- Bunny owns the reusable deterministic contract/profile vocabulary and golden-vector expectations.
- Wesley owns GraphQL SDL lowering, IR fidelity, and artifact generation.
- Runtime repos such as git-warp consume generated boundary artifacts and retain runtime truth locally.
Why this exists
git-warp issue git-stunts/git-warp#685 exposed a scalar/byte-fidelity gap before generated codecs can safely own canonical runtime boundary bytes. Bunny and Wesley now have companion issues for the profile and compiler sides:
Continuum should track the shared-family adoption side so this does not become a shadow vocabulary scattered through runtime repos.
Scope
Identify which Continuum family schemas should consume the deterministic profile first, likely around runtime-boundary/witnessed suffix exchange before broader families:
schemas/continuum-runtime-boundary-family.graphql
- receipt/witness family boundaries if they need byte-stable identifiers or envelopes
- any future optic / law / observer transport envelopes that must travel across runtimes
For each adopted field or envelope, Continuum should specify the semantic noun and proof posture while deferring byte representation mechanics to the Bunny/Wesley profile/toolchain.
Acceptance
- Continuum documents the ownership split: shared family semantics here, deterministic representation profile in Bunny, compiler lowering in Wesley, runtime truth in hosts.
- At least one shared boundary family identifies the fields/envelopes that require deterministic profile metadata.
- The adopted schema path can be compiled into generated artifacts without hand-maintained duplicate codecs in consuming repos.
- Continuum docs avoid implying that Continuum is a runtime, storage engine, or codec implementation.
Related
Context
Continuum owns shared boundary language and proof posture across the stack. It should not own runtime internals, storage, or compiler implementation, but its authored shared families need a deterministic representation profile so generated artifacts can travel without drift.
The active split is now:
Why this exists
git-warpissue git-stunts/git-warp#685 exposed a scalar/byte-fidelity gap before generated codecs can safely own canonical runtime boundary bytes. Bunny and Wesley now have companion issues for the profile and compiler sides:Continuum should track the shared-family adoption side so this does not become a shadow vocabulary scattered through runtime repos.
Scope
Identify which Continuum family schemas should consume the deterministic profile first, likely around runtime-boundary/witnessed suffix exchange before broader families:
schemas/continuum-runtime-boundary-family.graphqlFor each adopted field or envelope, Continuum should specify the semantic noun and proof posture while deferring byte representation mechanics to the Bunny/Wesley profile/toolchain.
Acceptance
Related