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/bad-code/PLATFORM_echo-wesley-gen-monolith.md
Original lane: bad-code
Original legend: PLATFORM
Original backlog card
echo-wesley-gen is a 2000-line monolith with mixed concerns
Status: bad code.
Where
crates/echo-wesley-gen/src/main.rs — single 2000+ line file.
Smell
One file emits:
- GraphQL → Rust type definitions (struct/enum)
- LE binary
Encode/Decode impls
- Op id constants +
OP_*_ARGS tables
- Footprint certificates + observer plan identities
RegistryProvider implementation + static REGISTRY__echo_wesley_generated helper module with Vars structs- EINT intent packers (
pack_*_intent, pack_*_intent_raw_vars)
- Optic dispatch/observe request builders
- Contract-host helpers (
*_contract_rule, *_contract_vars,
*_query_observer) gated by --contract-host
Every consumer pays the cost of every concern. Tests are slow (full
cargo run + smoke crate compile per case). Adding a new emit feature
means threading a new flag through eight code paths.
Why it matters
This file is also where the cross-language stable_op_id algorithm
lived until 2026-05-28. The fact that algorithm-as-source-of-truth
got buried in a code-emitter monolith is precisely the kind of smell
that delayed Wesley adoption.
Suggested split
echo-wesley-gen-types — struct/enum + Encode/Decode emitecho-wesley-gen-ops — op_id table + arg descriptors + RegistryProviderecho-wesley-gen-intent — EINT packers + optic helpersecho-wesley-gen-contract-host — gated host-side helpers- A thin top-level binary that composes them
Each compose-able piece can be tested in isolation. Algorithms shared
across targets (e.g., stable_op_id) migrate to wesley-core rather
than a vendored copy.
Surface when
Touching anything in this file. The next agent who has to add an emit
target will feel the pain; that's a good moment to extract.
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/bad-code/PLATFORM_echo-wesley-gen-monolith.mdOriginal lane:
bad-codeOriginal legend:
PLATFORMOriginal backlog card
echo-wesley-gen is a 2000-line monolith with mixed concerns
Status: bad code.
Where
crates/echo-wesley-gen/src/main.rs— single 2000+ line file.Smell
One file emits:
Encode/DecodeimplsOP_*_ARGStablesRegistryProviderimplementation + staticREGISTRY__echo_wesley_generatedhelper module withVarsstructspack_*_intent,pack_*_intent_raw_vars)*_contract_rule,*_contract_vars,*_query_observer) gated by--contract-hostEvery consumer pays the cost of every concern. Tests are slow (full
cargo run + smoke crate compile per case). Adding a new emit feature
means threading a new flag through eight code paths.
Why it matters
This file is also where the cross-language
stable_op_idalgorithmlived until 2026-05-28. The fact that algorithm-as-source-of-truth
got buried in a code-emitter monolith is precisely the kind of smell
that delayed Wesley adoption.
Suggested split
echo-wesley-gen-types— struct/enum + Encode/Decode emitecho-wesley-gen-ops— op_id table + arg descriptors + RegistryProviderecho-wesley-gen-intent— EINT packers + optic helpersecho-wesley-gen-contract-host— gated host-side helpersEach compose-able piece can be tested in isolation. Algorithms shared
across targets (e.g.,
stable_op_id) migrate towesley-coreratherthan a vendored copy.
Surface when
Touching anything in this file. The next agent who has to add an emit
target will feel the pain; that's a good moment to extract.