Echo and git-warp compatibility sanity check

[flyingrobots]

Migrated from Method backlog

This issue was created from a legacy filesystem backlog card. GitHub Issues are now the live work tracker; repository docs remain Method evidence.

Source backlog: docs/method/backlog/asap/KERNEL_echo-git-warp-compatibility-sanity-check.md
Original lane: asap
Original legend: KERNEL

Original backlog card

Echo and git-warp compatibility sanity check

A systematic review of where Echo and git-warp align, where they
diverge, and what needs to happen for the two substrates to share
a debugger, a protocol, and a schema compiler.

Context

Echo (Rust, deterministic simulation) and git-warp (JS, CRDT graph DB
on Git) are both WARP substrates. warp-ttd is the shared debugger.
Wesley is the shared schema compiler. But the two substrates evolved
independently and have diverged in ways that affect protocol design,
tooling, and future feature work (especially strands/braiding).

This item exists because a deep-dive in April 2026 surfaced
significant gaps that need deliberate alignment, not accidental drift.

Status: active coordination backlog. This is a compatibility map, not an
implementation spec; keep it aligned with current Echo runtime facts before
using it to drive cross-repo protocol work.

Triage update: keep this card, but sharpen the next pass around causal-history
interchange. The live question is no longer just whether Echo and git-warp can
share a debugger, protocol, and schema compiler; it is whether they can share
witnessed causal history through Continuum transport without making either
runtime subordinate to the other.

Where they align

Causal model

• Both use Lamport-style logical ticks (not wall-clock time)
• Both track per-writer provenance
• Both produce deterministic replay from identical inputs
• Both use receipts as structured provenance (applied/superseded/redundant)

Protocol surface

• Both are targets of the warp-ttd protocol
• Both are targets for Wesley-compiled protocol/schema surfaces. Echo currently
  has generated TTD protocol consumers plus local runtime schema fragments; not
  every runtime fragment is Wesley-generated yet.
• Both expose worldlines, lanes, and frame-indexed playback

Terminology

• Worldline, tick, receipt, observer, aperture — shared vocabulary
• warp-ttd's glossary enforces this (tested in CI)

Where they diverge

Convergence model

	Echo	git-warp
Strategy	Canonical order (one scheduler, one result)	CRDT convergence (OR-Set + LWW, commutative)
Multi-writer	Explicit sync protocol, scheduler-mediated	Native, automatic convergence
Merge	No CRDT merge API; base-worldline strand settlement exists	CRDT rules resolve automatically
Determinism	Non-negotiable — identical inputs produce identical outputs	Deterministic given same patches in same order

This is the fundamental divergence. Echo's determinism is
scheduler-enforced; git-warp's is CRDT-enforced. Both are
deterministic, but the mechanisms are incompatible.

Strands and braiding

	Echo	git-warp
Fork	ProvenanceStore::fork() plus StrandRegistry	createStrand() — pinned base observation + overlay writer
Strand type	First-class Strand with base ref, child worldline, writer heads, and support pins	Full: base observation, overlay, intent queue, evolution
Braid	Support pins exist; full braid presentation/collapse is not implemented	Read-only composition of support strands without collapse
Merge back	Base-worldline strand settlement exists; no general braid collapse	braidStrand() merges overlay into base graph
Compare	StrandSettlement::compare exists for base-worldline settlement	compareStrand() diffs strand vs any coordinate
Intent queue	Not present	Speculative intents queued but not committed

Echo now has initial strand lifecycle, support-pin, and base-worldline
settlement semantics. It still lacks git-warp's richer braid composition,
durable overlay, intent queue, and general comparison/collapse model.

Tick model

	Echo	git-warp
Tick identity	WorldlineTick (per-worldline) + GlobalTick (scheduler cycle)	Lamport clock (per-writer, monotonic)
Scheduler	First-class: states, modes, run IDs, work state	None — replay is on-demand
Playback	PlaybackCursor with 5 modes + SeekThen policy	Seek cache + named bookmarks

Ingress routing

	Echo	git-warp
Model	IngressTarget (3-variant union) + InboxPolicy (3-variant union)	Direct writer append
Head admission	HeadEligibility (Dormant/Admitted) + HeadDisposition (4 states)	No concept — all writers always active

Storage

	Echo	git-warp
Persistence	Runtime state is still in-memory/ephemeral; echo-cas exists but is not yet the runtime persistence layer	Git-native (durable, distributed)
Checkpoints	ProvenanceStore checkpoint interface (wormhole-ready)	Git snapshots after N patches
CAS	echo-cas crate exists; runtime persistence integration remains pending	git-cas (content-addressed blob store on Git)

Effect pipeline

	Echo	git-warp
Channels	Typed with policies (StrictSingle/Reduce/Log) + reducers	Effect entities (nodes with @warp/effect: prefix)
Delivery	Channel materialization + FinalizedChannel	EffectPipeline with externalization policy + sinks
Compliance	echo-ttd PolicyChecker with structured violations	No compliance concept

Protocol gaps (warp-ttd)

The warp-ttd protocol was shaped by git-warp. Seven specific gaps
exist for Echo (detailed in warp-ttd backlog item
PROTO_echo-runtime-schema-alignment):

1. Typed opaque identifiers — protocol uses String!, Echo uses
   32-byte hash-backed scalars
2. WorldlineTick vs GlobalTick — protocol collapses both to Int
3. Playback control — protocol has 3 mutations, Echo has 5 modes
   plus SeekThen
4. Ingress routing — not present in protocol at all
5. Head eligibility/disposition — not present
6. Scheduler introspection — not present
7. WriterHeadKey composite — flattened to single string

Schema compiler (Wesley)

Both substrates use Wesley for schema compilation, but:

• git-warp generates TypeScript types + Zod validators
• Echo generates Rust types via echo-wesley-gen and echo-ttd-gen
• warp-ttd has its own schema (warp-ttd-protocol.graphql)
• Echo has runtime schema fragments (schemas/runtime/artifact-a through d)
• These schemas are not yet coordinated

What needs to happen

Short term (coordinate, don't build)

• Verify whether Echo and git-warp can exchange witnessed causal history through
  Continuum transport while preserving each runtime's local admission law,
  receipt semantics, and storage model.
• Audit whether Echo's WASM ABI surface is sufficient for a warp-ttd
  host adapter (see PLATFORM_echo-ttd-host-adapter backlog item)
• Reconcile TTD protocol types — one schema, one source of truth
  (see PLATFORM_ttd-schema-reconciliation backlog item)
• Propose protocol extensions for Echo's richer runtime model
  (see warp-ttd backlog: PROTO_echo-runtime-schema-alignment)

Medium term (design decisions needed)

• Build on the live-basis strand semantics captured in
  docs/design/0010-live-basis-settlement-plan/design.md
• Design compliance reporting as a protocol extension
  (see KERNEL_compliance-protocol-envelope backlog item)
• Keep Echo's app-safe browser surface centered on warp-wasm, while
  warp-ttd owns debugger/browser delivery semantics.

Long term (requires strand/merge design)

• Implement Echo strands with base observation pinning
• Design canonical merge semantics (not CRDT convergence)
• Expose parallel execution counterfactuals through the debugger
• Cross-substrate debugging (debug Echo and git-warp in the same
  warp-ttd session)

[coderabbitai]

🔗 Related PRs

#326 - Echo contract hosting roadmap [merged]
#329 - Stack Witness 0001 fixture-backed jedit walking skeleton [merged]
#365 - feat(wesley): emit query observer host helpers [merged]
#373 - chore: remove legacy ttd-browser surface [merged]
#382 - 0024 universal LE binary codec + 0025 sessions design (design-only) [merged]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!