review-followup Wave 3: 설정 prefix app.* 통일(P20), Wave 3 종료

- 결정 16: `ca-skeleton.*`·`backend.graphql` 등 → `app.*`(APP_* env 키 이름 유지), selector 키 상수 하나(`CapabilitySelectors`),
  바인딩만 되던 `streaming-idle` 삭제
- `StrictSettingsBinder`가 env 오타까지 거부 — 모든 capability(graphql·websocket·mongo는 edge 제약으로 파일만 strict),
  httpclient 리플렉션 `HttpClientEnvironmentKeys` 삭제
- persistence-jpa가 `APP_DATASOURCE_PASSWORD` secret 요구를 기여(bootstrap이 JPA 스위치를 읽지 않음), Redis ACL fixture에서 pub/sub 권한 제거
- 결함 수정: SMTP 설정 env 이름 충돌(P20-new-1), `bindExcept`가 system env까지 filter(P20-new-2), graphql 빈 deployment-mode 메시지(P20-new-6)
- 통합: P30과 충돌 11건(P30의 명시적 bean 구조 유지), P30 코드의 옛 키를 app.*로, persistence-jpa 포맷
- 현황표: 소유자 결정 36(persistence-jpa 설정은 플랫폼 키만 app.persistence-jpa.*로, 공유 기능 트리 예외), 재개 문서에 일시 정지 상태

Wave 3 종료 검증: check · ci verifyDependencyLocks --warning-mode=fail(714 task) · mainCheck · architectureCheck
verifyPublicPathSnapshot :app-bootstrap:verifyEnvKeys configContractCheck · :app-bootstrap:integrationTest
:sample-portfolio:integrationTest :sample-portfolio:systemTest · integrationCheck · -p build-logic test ·
-p optional-platforms ci · :app-bootstrap:qualificationBootJar releaseCheck 통과

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This commit is contained in:
donghyeon-ka
2026-09-29 17:01:54 +09:00
co-authored by Claude Opus 5.5
parent 8bbf1bf01d
commit 27f23add08
295 changed files with 2284 additions and 1484 deletions
+7
View File
@@ -519,3 +519,10 @@ web 경로의 `W/`는 `src/adapter/inbound/web/src/main/java/dev/caskeleton/adap
| `PersistenceEntityScanCoverageTest.operationAndLiveEventAdaptersAreScanned` | `src/adapter/outbound/persistence-jpa/src/test/java/dev/caskeleton/adapter/outbound/persistence/config/PersistenceEntityScanCoverageTest.java` | `0ebb77b5` | 같은 이유(스캔 목록 고정) — `JpaAdapterBeansConfigTest.everyAdapterTheScanUsedToRegisterIsDeclaredAsABean`로 대체 |
| `MongoPlatformWiringTest.theMappingConfigurationCarriesThePlatformSwitch`·`thePlatformIsOffUnlessTheSwitchIsOn` | `src/adapter/outbound/persistence-mongo/src/test/java/dev/caskeleton/adapter/outbound/mongo/autoconfigure/MongoPlatformWiringTest.java` | `0ebb77b5` | 반복 스위치를 고정하던 테스트 — 반복 조건을 없애고 ArchUnit 두 규칙으로 대체 |
| `PortfolioEventContracts` | `src/sample-portfolio/src/main/java/dev/caskeleton/sample/portfolio/application/event/PortfolioEventContracts.java` | `0ebb77b5` | 공유 registry가 `event` → 모든 기능 역방향 의존 — `poster/PosterEventContracts`·`posterimage/PosterImageEventContracts`로 분할, `all()`은 테스트로(S6-#10a, `62a669c8`) |
| `HttpClientEnvironmentKeys`(설정 record 트리를 리플렉션으로 순회해 env 이름 역산·거부) | `src/adapter/outbound/httpclient/src/main/java/dev/caskeleton/adapter/outbound/httpclient/autoconfigure/HttpClientEnvironmentKeys.java` | `0ebb77b5` | S3-lead-2: env 오타 거부를 `StrictSettingsBinder`가 모든 capability에 대해 한다. manifest 도출은 `HttpClientPlatformEnvManifestTest`로 |
| `StrictSettingsBinder.bind(env, prefix, type, Predicate<ConfigurationPropertySource>)` overload | `src/adapter/outbound/support/.../StrictSettingsBinder.java` | `0ebb77b5` | httpclient 전용 우회였다. system env 제외가 기본 동작이 됨 |
| 설정 키 `app.httpclient.clients[N].timeout.streaming-idle`(`APP_HTTPCLIENT_CLIENTS_N_TIMEOUT_STREAMING_IDLE`)과 `TimeoutSettings.streamingIdle`·`HttpClientPlatformSettings.Timeout.streamingIdle` | `src/adapter/outbound/httpclient/.../profile/TimeoutSettings.java`, `docs/httpclient/env-fields.yaml` | `0ebb77b5` | P11-new-1: 바인딩만 되고 읽는 곳 없음(capability 규칙 3) |
| yml 키 `ca-skeleton.rate-limit.{limit,window,algorithm}` | `src/app-bootstrap/src/test/resources/application-test.yml`, `src/sample-portfolio/src/systemTest/resources/application-test.yml` | `0ebb77b5` | 읽는 코드 없음. `app.rate-limit` 아래로 옮기면 web `EdgeRateLimitTransportSettings` 트리에 섞인다 |
| `SecretSourceValidator`의 `RELATIONAL_PERSISTENCE_PROPERTY`·`isRelationalPersistenceOn()`(모듈 스위치 해석) | `src/app-bootstrap/src/main/java/dev/caskeleton/bootstrap/runtime/SecretSourceValidator.java` | `0ebb77b5` | P22-new-1: persistence-jpa가 기여 |
| 설정 키 `app.notification.platform.smtp.max-concurrency`(이름 변경 → `dispatch-concurrency`, env `APP_NOTIFICATION_PLATFORM_SMTP_DISPATCH_CONCURRENCY` 그대로) | `NotificationSmtpSettings.java`, `notification.yml` | `0ebb77b5` | P20-new-1 |
| Redis ACL `+@pubsub`·`&prod:*`(application 계정 2개) | `infra/redis-sdk/acl/*.acl` | `0ebb77b5` | P22-new-2 |
@@ -4,7 +4,7 @@ Every capability in `:grpc-advanced:*`, its grade, and what it would take to rai
`GrpcAdvancedSupportMatrix` is the machine-readable form; `GrpcAdvancedCapability.defaultGrade`
carries the same values.
All capabilities are off by default. Flags are `ca-skeleton.grpc.advanced.<capability>.enabled`.
All capabilities are off by default. Flags are `app.grpc.advanced.<capability>.enabled`.
## Grades
+3 -3
View File
@@ -14,8 +14,8 @@ exposes.
## Minimum to start
```yaml
ca-skeleton:
fileserver:
app:
fileserver-platform:
enabled: true
instance-id: ${HOSTNAME} # writer-lease owner; must be unique per node
storage:
@@ -47,7 +47,7 @@ Startup fails, rather than degrading, when any of these is missing or unsafe:
These roles decide the tier inside each operation. In front of it, every operation's use case
carries a `file:*` permission that the web adapter's method security checks against
`ca-skeleton.authz.role-permissions`; `security.yml` grants `file:read` to `ROLE_FILE_READ`,
`app.authz.role-permissions`; `security.yml` grants `file:read` to `ROLE_FILE_READ`,
`file:upload`, `file:delete`, `file:copy` and `file:move` to `ROLE_FILE_WRITE`, and `file:admin` to
`ROLE_FILE_ADMIN`. Renaming a tier role here means mapping the new name there as well, or the
caller is refused before the tier is consulted (`docs/fileserver/design-deviations.md` §5).
+1 -1
View File
@@ -198,7 +198,7 @@ append, finalize, cancel, the single-request upload), `file:delete`, `file:copy`
`ROLE_FILE_WRITE` and `ROLE_FILE_ADMIN` — the fileserver's own default tier roles — so a caller
holding a tier holds its permissions. A deployment that renames the tier roles, or supplies its own
`FileAccessPolicyPort` under other role names, maps those names in
`ca-skeleton.authz.role-permissions` too; `security.access-policy: unenforced` relaxes only the
`app.authz.role-permissions` too; `security.access-policy: unenforced` relaxes only the
tier and ownership check, never the permission gate.
The resumable transfers run on the transfer pool, and method security reads the security context
+4 -4
View File
@@ -6,8 +6,8 @@ The Fileserver ships off. Nothing is registered — no endpoint, no thread pool,
is enabled explicitly.
```yaml
ca-skeleton:
fileserver:
app:
fileserver-platform:
enabled: true
instance-id: ${HOSTNAME}
default-namespace: default
@@ -26,8 +26,8 @@ to an unkeyed digest, which would be reversible for an enumerable identifier spa
Each is a separate switch, and each defaults to off:
```yaml
ca-skeleton:
fileserver:
app:
fileserver-platform:
admin:
enabled: false # management plane; intended for a management port
tus:
@@ -99,7 +99,6 @@ materialise a nameless client in every deployment, which the aggregate validatio
| `response-header` | `2s` | Until final response headers |
| `read-idle` | `3s` | Between response chunks |
| `total-call` | `4s` | The whole logical call, including retry backoff |
| `streaming-idle` | `30s` | Silence on a long-lived stream |
`total-call` must not be shorter than `connect` or `response-header`; the validator emits
`INVALID_TIMEOUT_BUDGET` otherwise.
+3 -5
View File
@@ -1,8 +1,8 @@
# HTTP Client platform — Java field path to environment variable template.
#
# The SSOT is HttpClientPlatformSettings. HttpClientEnvironmentKeys derives this list from the
# record tree at runtime, HttpClientPlatformEnvManifestTest fails when the two disagree in either
# direction, and the platform refuses to start on an APP_HTTPCLIENT_ variable that is not here. So a
# The SSOT is HttpClientPlatformSettings. HttpClientPlatformEnvManifestTest derives this list from
# the record tree and fails when the two disagree in either direction, and the shared strict binder
# refuses to start on an APP_HTTPCLIENT_ variable that names no field. So a
# field added with no entry, an entry whose field was renamed, and a misspelled variable in a
# deployment are all failures rather than silence.
#
@@ -81,8 +81,6 @@ fields:
env: APP_HTTPCLIENT_CLIENTS_N_TIMEOUT_READ_IDLE
- field: clients[N].timeout.total-call
env: APP_HTTPCLIENT_CLIENTS_N_TIMEOUT_TOTAL_CALL
- field: clients[N].timeout.streaming-idle
env: APP_HTTPCLIENT_CLIENTS_N_TIMEOUT_STREAMING_IDLE
- field: clients[N].redirect.enabled
env: APP_HTTPCLIENT_CLIENTS_N_REDIRECT_ENABLED
- field: clients[N].redirect.max-hops
+1 -1
View File
@@ -28,7 +28,7 @@ producing-service identity다.
outbox는 canonical 하나다(2026-09-28). use case가 `IntegrationEventEncoderPort`로 만든
`ValidatedIntegrationEvent`를 `OutboxAppendPort`로 append하고, relay는 claim한 row의 exact envelope
bytes를 `IntegrationEventPublishPort`로 보낸다. 예전의 `ca-skeleton.outbox.canonical-transport-enabled`
bytes를 `IntegrationEventPublishPort`로 보낸다. 예전의 `app.outbox.canonical-transport-enabled`
gate와 legacy `MessageBroker → KafkaSender` 경로는 삭제됐다. 발행할 contract의 logical destination마다
아래 destination profile을 선언한다(템플릿의 durable operation event는 `durable-operations`).
+1 -1
View File
@@ -82,7 +82,7 @@ feature request:
| Tier | Capabilities | Enablement |
|---|---|---|
| Stable, auto-configured | Mapping and BSON converters, imperative execution, atomic update, bulk partial results, consistency profiles, retry/translation, operation budgets, observability, typed client profiles, startup validation, health | Beans registered when `ca-skeleton.persistence-mongo.enabled=true`. |
| Stable, auto-configured | Mapping and BSON converters, imperative execution, atomic update, bulk partial results, consistency profiles, retry/translation, operation budgets, observability, typed client profiles, startup validation, health | Beans registered when `app.persistence-mongo.enabled=true`. |
| Stable, library API | Transactions, causal sessions, optimistic revision, query builder and keyset pagination, aggregation executor, schema/index/validator manifests and diffs, TTL contract, GeoJSON, migration runner, admin plane, native capability gateway | The adopter's adapter leaf takes this leaf as `implementation` and constructs the type. |
| Advanced, library API | Sharding-aware query and shard admin, CSFLE, Queryable Encryption (equality/range), shared-collection tenancy policies | Same, plus a `MongoAdvancedCapabilityGuard` whose flags enable the capability. No configuration property. |
| Experimental, library API | Database-per-tenant client registry and migration coordinator, reshard approval | Same as Advanced; promotion additionally requires the evidence in [ADR-MONGO-ADV-001](../adr/ADR-MONGO-ADV-001-capability-promotion.md). |
+1 -1
View File
@@ -104,7 +104,7 @@ the three they achieved.
one thing — `AesGcmCallbackPayloadProtection`, which protects raw callback bodies — and by nothing on
the accept path. Demanding it would make a deployment supply a key that protects nothing while the
payload it appears to be about stays in plaintext. The repository already has one defect of that
exact shape: `backend.graphql.cursor.key-ids` (key removed 2026-09-25), which production refuses to start without and which no
exact shape: `app.graphql.cursor.key-ids` (key removed 2026-09-25), which production refuses to start without and which no
code signs a cursor with (GQL-INT-003). Adding a second would make the pattern a habit.
## Consequence
+3 -3
View File
@@ -1,6 +1,6 @@
# Configuration reference
The notification delivery platform binds under `ca-skeleton.notification.platform`. The tree lives in
The notification delivery platform binds under `app.notification.platform`. The tree lives in
`src/composition/notification-platform/src/main/resources/config/module-defaults/notification.yml`,
which the support leaf's `ModuleDefaultsEnvironmentPostProcessor` loads at the lowest precedence,
so an application without the module carries none of it. The module is in no default composition: a
@@ -60,7 +60,7 @@ nonce and tag. A larger configured value would pass every check above the databa
## Provider profiles
Profiles are a map under `providers`, keyed by profile id. Arbitrary deployment-chosen ids can be
supplied as YAML or through Spring relaxed binding (`CA_SKELETON_NOTIFICATION_PLATFORM_PROVIDERS_<ID>_<FIELD>`).
supplied as YAML or through Spring relaxed binding (`APP_NOTIFICATION_PLATFORM_PROVIDERS_<ID>_<FIELD>`).
The shipped reference profiles `smtp` and `ses` additionally expose explicit `APP_NOTIFICATION_PLATFORM_SMTP_*`
and `APP_NOTIFICATION_PLATFORM_SES_*` aliases so operators do not have to invent binding names.
@@ -135,7 +135,7 @@ already owns them and a second spelling would be a second thing to keep in step.
| Property | Environment variable | Default | Bound |
|---|---|---|---|
| `smtp.sender-identity` | `APP_NOTIFICATION_PLATFORM_SMTP_SENDER_IDENTITY` | `no-reply@example.invalid` | address |
| `smtp.max-concurrency` | `APP_NOTIFICATION_PLATFORM_SMTP_DISPATCH_CONCURRENCY` | `4` | positive |
| `smtp.dispatch-concurrency` | `APP_NOTIFICATION_PLATFORM_SMTP_DISPATCH_CONCURRENCY` | `4` | positive |
SMTP transport configuration has one owner: Spring Mail. The actual `JavaMailSenderImpl` must use required STARTTLS (`mail.smtp.starttls.enable=true` plus `mail.smtp.starttls.required=true`) or implicit TLS (`mail.smtp.ssl.enable=true`). It must also have positive `mail.smtp.connectiontimeout`, `mail.smtp.timeout`, and `mail.smtp.writetimeout` values. SMTP assembly fails before serving traffic when those properties are absent or invalid.
+33 -33
View File
@@ -337,7 +337,7 @@ env_keys:
- name: APP_PERSISTENCE_VENDOR
# source: persistence vendor selector used by application-local.yml and bootRunH2.
# PostgreSQL is the shipped default; H2 is an explicit developer-only selection.
property: ca-skeleton.persistence.vendor
property: app.persistence.vendor
type: enum
default: postgresql
allowed_values: [postgresql, h2]
@@ -350,7 +350,7 @@ env_keys:
required_test: jpa-contract:local-vendor-parity
- name: APP_REALTIME_REPLAY_ENABLED
property: ca-skeleton.realtime.replay.enabled
property: app.realtime.replay.enabled
owner_module: adapter-outbound-persistence-jpa
type: boolean
default: false
@@ -364,7 +364,7 @@ env_keys:
required_test: jpa-contract:realtime-replay-env
- name: APP_REALTIME_REPLAY_RETENTION
property: ca-skeleton.realtime.replay.retention
property: app.realtime.replay.retention
owner_module: adapter-outbound-persistence-jpa
type: duration
default: 1h
@@ -1596,7 +1596,7 @@ env_keys:
required_test: rate-limit-contract:cleanup-grace-bounded
- name: APP_IDEMPOTENCY_TTL
# ca-skeleton.capabilities.idempotency.replay-ttl: how long a completed response stays
# app.capabilities.idempotency.replay-ttl: how long a completed response stays
# replayable. One value for every provider; IdempotencySettings refuses more than 72h at startup.
type: duration
default: 24h
@@ -1610,7 +1610,7 @@ env_keys:
required_test: idempotency-contract:ttl-applied
- name: APP_IDEMPOTENCY_PROVIDER
# ca-skeleton.capabilities.idempotency.provider. redis and postgresql each contribute only the
# app.capabilities.idempotency.provider. redis and postgresql each contribute only the
# store; app-bootstrap builds the one IdempotencyExecutor and refuses startup unless exactly one
# store exists. postgresql uses the primary data source, applies its Flyway streams at startup
# and refuses a non-PostgreSQL data source, so bootRunH2 runs with disabled or redis. Selection
@@ -2350,7 +2350,7 @@ env_keys:
# DEPRECATED 2026-08-10: this was a second Redis master switch, and the settings class that
# bound it was removed with the previous Redis generation, so the key reached nothing.
# APP_REDIS_ENABLED is the sole global activation authority; the cache role is selected by
# ca-skeleton.capabilities.cache.bindings.default. The row is kept rather than deleted so a
# app.capabilities.cache.bindings.default. The row is kept rather than deleted so a
# deployment still setting this key can be told what replaced it.
deprecated_orphaned: true
deprecated_alias_for: APP_REDIS_ENABLED
@@ -3682,7 +3682,7 @@ env_keys:
required_test: async-contract:executor-queue-bounded
# === Notification delivery platform (NTF-025 — configuration surface) ===
# The tree exists in application.yml as ca-skeleton.notification.platform, disabled by
# The tree exists in application.yml as app.notification.platform, disabled by
# default. Every key carries an inline default so a deployment that leaves the platform off
# supplies nothing. Reference: docs/notification/configuration.md.
@@ -3816,7 +3816,7 @@ env_keys:
allowed_values: null
classification: sensitive-config
required: false
required_when: ca-skeleton.persistence-mongo.enabled=true AND no active-profile uri-secret
required_when: app.persistence-mongo.enabled=true AND no active-profile uri-secret
reload_policy: restart-only
owner_branch: persistence-mongo-runtime-remediation
validation: mongodb_uri_when_selected_as_connection_source
@@ -3934,8 +3934,8 @@ env_keys:
compatibility_impact: behavior-change
required_test: adapter-contract:graphql-disabled-safe
- name: APP_GRAPHQL_DEPLOYMENT_MODE
# source: five-adapter-runtime-remediation §5.1 / GQL-INT-002 — replaces backend.graphql.production
# and backend.graphql.environment, which defaulted to production=false with
# source: five-adapter-runtime-remediation §5.1 / GQL-INT-002 — replaces app.graphql.production
# and app.graphql.environment, which defaulted to production=false with
# environment=PRODUCTION_PUBLIC and let anonymous-principal and allow-by-default authorization
# read one axis while the other claimed production.
type: enum
@@ -3950,7 +3950,7 @@ env_keys:
compatibility_impact: behavior-change
required_test: adapter-contract:graphql-deployment-mode
- name: APP_GRAPHQL_VALIDATION_POLICY_VERSION
property: backend.graphql.validation-policy-version
property: app.graphql.validation-policy-version
owner_module: adapter-inbound-graphql
type: string
default: v1
@@ -3965,7 +3965,7 @@ env_keys:
required_test: graphql-contract:wired-operator-settings
- name: APP_GRAPHQL_OBSERVED_OPERATION_NAMES
property: backend.graphql.observed-operation-names
property: app.graphql.observed-operation-names
owner_module: adapter-inbound-graphql
type: csv
default: ""
@@ -3980,7 +3980,7 @@ env_keys:
required_test: graphql-contract:wired-operator-settings
- name: APP_GRAPHQL_GRAPHIQL_ENABLED
property: backend.graphql.console.graphiql-enabled
property: app.graphql.console.graphiql-enabled
owner_module: adapter-inbound-graphql
type: boolean
default: false
@@ -3995,7 +3995,7 @@ env_keys:
required_test: graphql-contract:wired-operator-settings
- name: APP_GRAPHQL_INTROSPECTION_ENABLED
property: backend.graphql.console.introspection-enabled
property: app.graphql.console.introspection-enabled
owner_module: adapter-inbound-graphql
type: boolean
default: false
@@ -4010,7 +4010,7 @@ env_keys:
required_test: graphql-contract:wired-operator-settings
- name: APP_GRAPHQL_MAXIMUM_PAGE_SIZE
property: backend.graphql.limits.maximum-page-size
property: app.graphql.limits.maximum-page-size
owner_module: adapter-inbound-graphql
type: integer
default: 100
@@ -4025,7 +4025,7 @@ env_keys:
required_test: graphql-contract:wired-operator-settings
- name: APP_GRAPHQL_MAXIMUM_COMPLEXITY
property: backend.graphql.limits.maximum-complexity
property: app.graphql.limits.maximum-complexity
owner_module: adapter-inbound-graphql
type: integer
default: 10000
@@ -4040,7 +4040,7 @@ env_keys:
required_test: graphql-contract:wired-operator-settings
- name: APP_GRAPHQL_PREPARSED_CACHE_ENTRIES
property: backend.graphql.limits.preparsed-cache-entries
property: app.graphql.limits.preparsed-cache-entries
owner_module: adapter-inbound-graphql
type: integer
default: 1000
@@ -4055,7 +4055,7 @@ env_keys:
required_test: graphql-contract:wired-operator-settings
- name: APP_GRAPHQL_PREPARSED_CACHE_WEIGHT
property: backend.graphql.limits.preparsed-cache-weight
property: app.graphql.limits.preparsed-cache-weight
owner_module: adapter-inbound-graphql
type: integer
default: 10000000
@@ -4070,7 +4070,7 @@ env_keys:
required_test: graphql-contract:wired-operator-settings
- name: APP_GRAPHQL_PREPARSED_CACHE_EXPIRE_AFTER_ACCESS
property: backend.graphql.limits.preparsed-cache-expire-after-access
property: app.graphql.limits.preparsed-cache-expire-after-access
owner_module: adapter-inbound-graphql
type: duration
default: 30m
@@ -4147,7 +4147,7 @@ env_keys:
- name: APP_NOTIFICATION_PLATFORM_CONTACT_ENCRYPTION_KEY
# source: NTF-INT-007 — Encrypts recipient contact points at rest — addresses and phone numbers.
# Full row owned by secrets-classification.yaml. Bound by
# ca-skeleton.notification.platform.secrets.contact-encryption-key; the platform decodes it at startup and
# app.notification.platform.secrets.contact-encryption-key; the platform decodes it at startup and
# refuses to boot if it is blank, shorter than 32 bytes, or equal to another purpose's key.
type: string
default: null
@@ -4163,7 +4163,7 @@ env_keys:
- name: APP_NOTIFICATION_PLATFORM_CONTACT_LOOKUP_HMAC_KEY
# source: NTF-INT-007 — Blind index over contact points, so a lookup never needs the plaintext.
# Full row owned by secrets-classification.yaml. Bound by
# ca-skeleton.notification.platform.secrets.contact-lookup-hmac-key; the platform decodes it at startup and
# app.notification.platform.secrets.contact-lookup-hmac-key; the platform decodes it at startup and
# refuses to boot if it is blank, shorter than 32 bytes, or equal to another purpose's key.
type: string
default: null
@@ -4179,7 +4179,7 @@ env_keys:
- name: APP_NOTIFICATION_PLATFORM_CALLBACK_SIGNING_KEY
# source: NTF-INT-007 — Signs the callback URLs a provider posts delivery outcomes back to.
# Full row owned by secrets-classification.yaml. Bound by
# ca-skeleton.notification.platform.secrets.callback-signing-key; the platform decodes it at startup and
# app.notification.platform.secrets.callback-signing-key; the platform decodes it at startup and
# refuses to boot if it is blank, shorter than 32 bytes, or equal to another purpose's key.
type: string
default: null
@@ -4195,7 +4195,7 @@ env_keys:
- name: APP_NOTIFICATION_PLATFORM_PROVIDER_CREDENTIAL_KEY
# source: NTF-INT-007 — Encrypts stored provider credentials, which are themselves secrets.
# Full row owned by secrets-classification.yaml. Bound by
# ca-skeleton.notification.platform.secrets.provider-credential-key; the platform decodes it at startup and
# app.notification.platform.secrets.provider-credential-key; the platform decodes it at startup and
# refuses to boot if it is blank, shorter than 32 bytes, or equal to another purpose's key.
type: string
default: null
@@ -4211,7 +4211,7 @@ env_keys:
- name: APP_NOTIFICATION_PLATFORM_PAYLOAD_ENCRYPTION_KEY
# source: NTF-INT-007 — Encrypts notification variables and retained callback bodies at rest.
# Full row owned by secrets-classification.yaml. Bound by
# ca-skeleton.notification.platform.secrets.payload-encryption-key; the platform decodes it at startup and
# app.notification.platform.secrets.payload-encryption-key; the platform decodes it at startup and
# refuses to boot if it is blank, shorter than 32 bytes, or equal to another purpose's key.
type: string
default: null
@@ -4227,7 +4227,7 @@ env_keys:
- name: APP_NOTIFICATION_PLATFORM_PROVIDER_REQUEST_LOOKUP_HMAC_KEY
# source: NTF-INT-007 — Keyed hash of provider request ids, which are provider-side identifiers.
# Full row owned by secrets-classification.yaml. Bound by
# ca-skeleton.notification.platform.secrets.provider-request-lookup-hmac-key; the platform decodes it at startup and
# app.notification.platform.secrets.provider-request-lookup-hmac-key; the platform decodes it at startup and
# refuses to boot if it is blank, shorter than 32 bytes, or equal to another purpose's key.
type: string
default: null
@@ -4243,7 +4243,7 @@ env_keys:
- name: APP_NOTIFICATION_PLATFORM_CALLBACK_FINGERPRINT_HMAC_KEY
# source: NTF-INT-007 — Keyed fingerprint of callback bodies, used to detect replays.
# Full row owned by secrets-classification.yaml. Bound by
# ca-skeleton.notification.platform.secrets.callback-fingerprint-hmac-key; the platform decodes it at startup and
# app.notification.platform.secrets.callback-fingerprint-hmac-key; the platform decodes it at startup and
# refuses to boot if it is blank, shorter than 32 bytes, or equal to another purpose's key.
type: string
default: null
@@ -4260,7 +4260,7 @@ env_keys:
# source: NTF-INT-007 — the id written into every envelope the CONTACT_ENCRYPTION key produces.
# An identifier, not key material, so it is public-config; the material itself is the
# APP_NOTIFICATION_PLATFORM_CONTACT_ENCRYPTION_KEY row above. Bound into
# ca-skeleton.notification.platform.secrets.active-key-ids.CONTACT_ENCRYPTION.
# app.notification.platform.secrets.active-key-ids.CONTACT_ENCRYPTION.
type: string
default: null
allowed_values: null
@@ -4276,7 +4276,7 @@ env_keys:
# source: NTF-INT-007 — the id written into every envelope the CONTACT_LOOKUP_HMAC key produces.
# An identifier, not key material, so it is public-config; the material itself is the
# APP_NOTIFICATION_PLATFORM_CONTACT_LOOKUP_HMAC_KEY row above. Bound into
# ca-skeleton.notification.platform.secrets.active-key-ids.CONTACT_LOOKUP_HMAC.
# app.notification.platform.secrets.active-key-ids.CONTACT_LOOKUP_HMAC.
type: string
default: null
allowed_values: null
@@ -4292,7 +4292,7 @@ env_keys:
# source: NTF-INT-007 — the id written into every envelope the CALLBACK_SIGNING key produces.
# An identifier, not key material, so it is public-config; the material itself is the
# APP_NOTIFICATION_PLATFORM_CALLBACK_SIGNING_KEY row above. Bound into
# ca-skeleton.notification.platform.secrets.active-key-ids.CALLBACK_SIGNING.
# app.notification.platform.secrets.active-key-ids.CALLBACK_SIGNING.
type: string
default: null
allowed_values: null
@@ -4308,7 +4308,7 @@ env_keys:
# source: NTF-INT-007 — the id written into every envelope the PROVIDER_CREDENTIAL key produces.
# An identifier, not key material, so it is public-config; the material itself is the
# APP_NOTIFICATION_PLATFORM_PROVIDER_CREDENTIAL_KEY row above. Bound into
# ca-skeleton.notification.platform.secrets.active-key-ids.PROVIDER_CREDENTIAL.
# app.notification.platform.secrets.active-key-ids.PROVIDER_CREDENTIAL.
type: string
default: null
allowed_values: null
@@ -4324,7 +4324,7 @@ env_keys:
# source: NTF-INT-007 — the id written into every envelope the PAYLOAD_ENCRYPTION key produces.
# An identifier, not key material, so it is public-config; the material itself is the
# APP_NOTIFICATION_PLATFORM_PAYLOAD_ENCRYPTION_KEY row above. Bound into
# ca-skeleton.notification.platform.secrets.active-key-ids.PAYLOAD_ENCRYPTION.
# app.notification.platform.secrets.active-key-ids.PAYLOAD_ENCRYPTION.
type: string
default: null
allowed_values: null
@@ -4340,7 +4340,7 @@ env_keys:
# source: NTF-INT-007 — the id written into every envelope the PROVIDER_REQUEST_LOOKUP_HMAC key produces.
# An identifier, not key material, so it is public-config; the material itself is the
# APP_NOTIFICATION_PLATFORM_PROVIDER_REQUEST_LOOKUP_HMAC_KEY row above. Bound into
# ca-skeleton.notification.platform.secrets.active-key-ids.PROVIDER_REQUEST_LOOKUP_HMAC.
# app.notification.platform.secrets.active-key-ids.PROVIDER_REQUEST_LOOKUP_HMAC.
type: string
default: null
allowed_values: null
@@ -4356,7 +4356,7 @@ env_keys:
# source: NTF-INT-007 — the id written into every envelope the CALLBACK_FINGERPRINT_HMAC key produces.
# An identifier, not key material, so it is public-config; the material itself is the
# APP_NOTIFICATION_PLATFORM_CALLBACK_FINGERPRINT_HMAC_KEY row above. Bound into
# ca-skeleton.notification.platform.secrets.active-key-ids.CALLBACK_FINGERPRINT_HMAC.
# app.notification.platform.secrets.active-key-ids.CALLBACK_FINGERPRINT_HMAC.
type: string
default: null
allowed_values: null
+18 -15
View File
@@ -24,6 +24,9 @@ secrets:
# "DB credential은 dual-bind 60s" + "__LOCAL_DEV_FAKE_DB_PASSWORD" 예시
classification: secret
source: secret-manager
# Contributed by persistence-jpa (JpaStartupAutoConfiguration#jpaProductionSecrets) while the
# module is on; the composition root's guard no longer reads the module switch (P22-new-1).
required_when: app.persistence-jpa.enabled=true
rotation_policy: dual-bind-60s
prod_default: null
dev_sentinel_prefix: __LOCAL_DEV_
@@ -55,7 +58,7 @@ secrets:
rotation_policy: restart-only
prod_default: null
# Required with app.redis.enabled=true and this role selected; checked by adapter-outbound-cache-redis RedisProductionSecrets.
required_when: ca-skeleton.capabilities.cache.bindings.default=redis
required_when: app.capabilities.cache.bindings.default=redis
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: feature-secrets-config-source-contract
masking_rule: full
@@ -81,7 +84,7 @@ secrets:
rotation_policy: dual-read-restart-only
prod_default: null
# Required with app.redis.enabled=true and this role selected; checked by adapter-outbound-cache-redis RedisProductionSecrets.
required_when: ca-skeleton.capabilities.cache.bindings.default=redis
required_when: app.capabilities.cache.bindings.default=redis
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: codex-phase-a-ci-recovery
masking_rule: full
@@ -95,7 +98,7 @@ secrets:
rotation_policy: restart-only
prod_default: null
# Required with app.redis.enabled=true and this role selected; checked by adapter-outbound-cache-redis RedisProductionSecrets.
required_when: ca-skeleton.capabilities.rate-limit.provider=redis
required_when: app.capabilities.rate-limit.provider=redis
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: redis-distributed-rate-limit
masking_rule: full
@@ -121,7 +124,7 @@ secrets:
rotation_policy: dual-read-restart-only
prod_default: null
# Required with app.redis.enabled=true and this role selected; checked by adapter-outbound-cache-redis RedisProductionSecrets.
required_when: ca-skeleton.capabilities.rate-limit.provider=redis
required_when: app.capabilities.rate-limit.provider=redis
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: redis-distributed-rate-limit
masking_rule: full
@@ -135,7 +138,7 @@ secrets:
rotation_policy: restart-only
prod_default: null
# Required with app.redis.enabled=true and this role selected; checked by adapter-outbound-cache-redis RedisProductionSecrets.
required_when: ca-skeleton.security.auth-mode=redis-session
required_when: app.security.auth-mode=redis-session
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: redis-production-capability
masking_rule: full
@@ -161,7 +164,7 @@ secrets:
rotation_policy: dual-read-restart-only
prod_default: null
# Required with app.redis.enabled=true and this role selected; checked by adapter-outbound-cache-redis RedisProductionSecrets.
required_when: ca-skeleton.security.auth-mode=redis-session
required_when: app.security.auth-mode=redis-session
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: redis-production-capability
masking_rule: full
@@ -175,7 +178,7 @@ secrets:
rotation_policy: cold-cutover-restart-only
prod_default: null
# Required with app.redis.enabled=true and this role selected; checked by adapter-outbound-cache-redis RedisProductionSecrets.
required_when: ca-skeleton.capabilities.idempotency.provider=redis
required_when: app.capabilities.idempotency.provider=redis
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: redis-production-capability-completion
masking_rule: full
@@ -189,7 +192,7 @@ secrets:
rotation_policy: cold-cutover-restart-only
prod_default: null
# Required with app.redis.enabled=true and this role selected; checked by adapter-outbound-cache-redis RedisProductionSecrets.
required_when: ca-skeleton.capabilities.lease.provider=redis
required_when: app.capabilities.lease.provider=redis
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: redis-production-capability-completion
masking_rule: full
@@ -222,7 +225,7 @@ secrets:
source: secret-manager
rotation_policy: dual-read-restart-only
prod_default: null
required_when: ca-skeleton.notification.platform.enabled=true
required_when: app.notification.platform.enabled=true
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: worktree-notification-platform
masking_rule: full
@@ -238,7 +241,7 @@ secrets:
source: secret-manager
rotation_policy: dual-read-restart-only
prod_default: null
required_when: ca-skeleton.notification.platform.enabled=true
required_when: app.notification.platform.enabled=true
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: worktree-notification-platform
masking_rule: full
@@ -254,7 +257,7 @@ secrets:
source: secret-manager
rotation_policy: dual-read-restart-only
prod_default: null
required_when: ca-skeleton.notification.platform.enabled=true
required_when: app.notification.platform.enabled=true
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: worktree-notification-platform
masking_rule: full
@@ -270,7 +273,7 @@ secrets:
source: secret-manager
rotation_policy: dual-read-restart-only
prod_default: null
required_when: ca-skeleton.notification.platform.enabled=true
required_when: app.notification.platform.enabled=true
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: worktree-notification-platform
masking_rule: full
@@ -286,7 +289,7 @@ secrets:
source: secret-manager
rotation_policy: dual-read-restart-only
prod_default: null
required_when: ca-skeleton.notification.platform.enabled=true
required_when: app.notification.platform.enabled=true
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: worktree-notification-platform
masking_rule: full
@@ -302,7 +305,7 @@ secrets:
source: secret-manager
rotation_policy: dual-read-restart-only
prod_default: null
required_when: ca-skeleton.notification.platform.enabled=true
required_when: app.notification.platform.enabled=true
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: worktree-notification-platform
masking_rule: full
@@ -318,7 +321,7 @@ secrets:
source: secret-manager
rotation_policy: dual-read-restart-only
prod_default: null
required_when: ca-skeleton.notification.platform.enabled=true
required_when: app.notification.platform.enabled=true
dev_sentinel_prefix: __LOCAL_DEV_
owner_branch: worktree-notification-platform
masking_rule: full
@@ -0,0 +1,99 @@
# W3-P20 — 설정 prefix `app.*` 통일(결정 16) + strict 바인딩
worktree: `.claude/worktrees/agent-a79f4feba1997eb73` (기준 HEAD `0ebb77b5`)
## 설정 prefix 이전 → 이후
env 키(`APP_*`) 이름은 하나도 바뀌지 않았다. 바뀐 것은 env가 채우는 설정 트리뿐이다.
| 이전 | 이후 | 소유(읽는 곳) |
|---|---|---|
| `ca-skeleton.async.executor` | `app.async.executor` | app-bootstrap `AsyncExecutorSettings` |
| `ca-skeleton.bootstrap` | `app.bootstrap` | app-bootstrap `BootstrapSettings` |
| `ca-skeleton.logging` | `app.logging` | app-bootstrap `LoggingSettings` |
| `ca-skeleton.runtime` | `app.runtime` | app-bootstrap `RuntimeSafetySettings`, persistence-jpa `migration-on-startup` |
| `ca-skeleton.secret-source` | `app.secret-source` | app-bootstrap `SecretSourceSettings` |
| `ca-skeleton.security` | `app.security` | app-bootstrap `AuthenticationModeSettings`, web `SecuritySettings` |
| `ca-skeleton.authz` | `app.authz` | web `RolePermissionPolicy` |
| `ca-skeleton.cors` | `app.cors` | web `CorsSettings` |
| `ca-skeleton.presentation` | `app.presentation` | web `PresentationSettings` |
| `backend.web.mvc` | `app.web.mvc` | web `WebMvcPlatformSettings` |
| `backend.web.advanced.mvc-virtual-threads` | `app.web.advanced.mvc-virtual-threads` | web-advanced `VirtualThreadSettings` |
| `backend.graphql` | `app.graphql` | graphql `GraphQlPlatformSettings` |
| `ca-skeleton.grpc.server` | `app.grpc.server` | grpc `GrpcServerSettings` |
| `ca-skeleton.grpc.platform` | `app.grpc.platform` | grpc starter `GrpcPlatformProperties` (부모 `app.grpc`에 바인딩하는 것은 없다 — P5 메모) |
| `ca-skeleton.grpc.advanced.<flag>.enabled` | `app.grpc.advanced.<flag>.enabled` | grpc-advanced `GrpcAdvancedCapability` |
| `ca-skeleton.capabilities.{cache,rate-limit,lease,idempotency}` | `app.capabilities.{…}` | cache-redis `RedisCapabilitySettings`, composition:idempotency `IdempotencySettings`, persistence-jpa idempotency |
| `ca-skeleton.notification.platform(.smtp,.secrets)` | `app.notification.platform(…)` | notification `NotificationPlatformSettings`·`NotificationSmtpSettings`·secrets |
| `ca-skeleton.persistence.vendor` | `app.persistence.vendor` | persistence-jpa `PersistenceVendorSettings` |
| `ca-skeleton.persistence-jpa.enabled` | `app.persistence-jpa.enabled` | persistence-jpa 스위치 |
| `ca-skeleton.lock` | `app.lock` | persistence-jpa `LockSettings` |
| `ca-skeleton.outbox` | `app.outbox` | composition:outbox-relay `OutboxSettings` |
| `ca-skeleton.realtime.replay` | `app.realtime.replay` | persistence-jpa `JpaLiveEventReplay` |
| `ca-skeleton.persistence-mongo(.platform)` | `app.persistence-mongo(…)` | persistence-mongo `MongoPersistenceSettings`·`MongoPlatformSettings` |
| `ca-skeleton.privacy` | `app.privacy` | composition:service-runtime `PrivacySettings` |
| `ca-skeleton.tracing` | `app.tracing` | composition:service-runtime `TracingSettings` |
| `ca-skeleton.rate-limit.{limit,window,algorithm}` (bootstrap·sample test yml) | 삭제 | 읽는 코드 없음. `ca-skeleton.rate-limit.{enabled,…}`(sample)는 `app.rate-limit.*`가 되어 web `EdgeRateLimitTransportSettings`가 읽는 키와 합쳐졌다(값 동일, enabled=false) |
| 그대로 | `app.redis`, `app.httpclient`, `app.fileserver-platform`, `app.object-storage`, `app.websocket`, `app.notification`(R0 gate), `app.rate-limit`, `app.jpa-platform`, `app.messaging`(vendored), `app.poster-image` | |
| 그대로(설정 키가 아님) | `ca-skeleton.messaging.{schema-set,contract-catalog,contract-descriptor,partition-key}.v1`(hash domain), `ca-skeleton.grpc.authenticated-caller`(gRPC Context key), `logs/ca-skeleton.json`, metric `backend.redis.*`, vendored messaging의 superseded `backend.messaging` 거부 | |
## strict 바인딩 현황 (X2-P1-3c)
| 설정 | prefix | 방식 |
|---|---|---|
| composition:fileserver-platform, httpclient, composition:outbox-relay, composition:idempotency | 기존 | `StrictSettingsBinder`(이번에 env 검사 일반화) |
| persistence-jpa `PersistenceVendorSettings`·`LockSettings`·`JpaSafetySettings` | `app.persistence`·`app.lock`·`app.jpa-platform` | `StrictSettingsBinder`(게이트된 설정의 `@Bean`) |
| cache-redis `RedisCapabilitySettings` | `app.capabilities` | `StrictSettingsBinder.bindExcept`(idempotency lifecycle 키 제외) |
| notification `NotificationRoutesSettings`·`NotificationPlatformSettings`·secrets·smtp | `app.notification`(platform·R0 gate 제외)·`app.notification.platform`(smtp·secrets·default-locale 제외)·`.secrets`·`.smtp` | `StrictSettingsBinder` |
| messaging bridge `MessagingSettings` | `app.messaging`(vendored 최상위 항목 제외) | `StrictSettingsBinder.bindExcept` |
| service-runtime `PrivacySettings`·`TracingSettings` | `app.privacy`·`app.tracing` | `StrictSettingsBinder` |
| objectstorage `ObjectStorageCapabilitySettings` | `app.object-storage`(scan-maintenance 제외) | `StrictSettingsBinder` + registry support edge 추가 |
| graphql `GraphQlPlatformSettings` | `app.graphql` | `ignoreUnknownFields = false`(inbound → outbound support edge 불가). 스위치 `enabled`를 component로 추가 |
| websocket `WebSocketSettings` | `app.websocket` | `ignoreUnknownFields = false`(같은 이유) |
| persistence-mongo `MongoPlatformSettings` | `app.persistence-mongo.platform` | `ignoreUnknownFields = false`(closure에 support 없음). 부모 `app.persistence-mongo`(스위치 하나)는 하위 트리 때문에 엄격하게 하지 않음 |
| grpc `GrpcServerSettings`·`GrpcPlatformProperties` | `app.grpc.server`·`app.grpc.platform` | 기존 `ignoreUnknownFields = false`, 부모 `app.grpc`에 바인딩 없음 |
## 행 결과
| ID | 결과 | 근거 |
|---|---|---|
| S6-#14 | 완료 | 결정 16: 설정 prefix를 `app.*` 하나로(위 표). `711a3968`·`fabb611e`. env 키 이름 변경 없음. `ConfigurationNamespaceContractTest`: 허용 root는 `app` 하나(`shippedConfigurationDeclaresOnlyRecordedRoots`), 모든 leaf의 `@ConfigurationProperties`·`String PREFIX` 상수가 `app.` 아래(`everyLeafPrefixSitsUnderTheApplicationRoot`, 새 테스트) — `src/app-bootstrap/src/test/java/dev/caskeleton/bootstrap/contract/ConfigurationNamespaceContractTest.java`. yml root 병합 4곳(cache-redis·notification module-defaults, sample `application.yml`·systemTest yml). "capability" 용어 정리는 이번 범위 밖(키 이름만) |
| S3-lead-4 | 완료 | `711a3968`: `src/adapter/outbound/support/src/main/java/dev/caskeleton/adapter/outbound/support/CapabilitySelectors.java`(ROOT·CACHE_DEFAULT_BINDING·IDEMPOTENCY_PROVIDER·RATE_LIMIT_PROVIDER·LEASE_PROVIDER). 사용: cache-redis `RedisCacheConfig`·`RedisCorrectnessRoleSelectors`·`RedisIdempotencyConfig`·`RedisLeaseConfig`·`RedisRateLimitConfig`·`RedisCapabilitySettings.bind`, persistence-jpa `PostgreSqlIdempotencyStoreConfig`·`IdempotencyReaper`·`IdempotencySchemaStream`, composition:idempotency `IdempotencyAutoConfiguration`. 현황표 방향(멱등성은 composition 모듈에)과 다르게 support에 둔 이유: persistence-jpa·cache-redis는 composition:idempotency에 의존할 수 없고(registry), 셋 다 support에 이미 의존한다. main의 `app.capabilities.*.provider` 문자열 리터럴 0건(`git grep '"app.capabilities' -- '*/src/main/*'`는 메시지·주석뿐) |
| P11-new-1 | 완료 | `711a3968`: 읽지 않던 `clients[N].timeout.streaming-idle` 키 삭제 — `TimeoutSettings`·`HttpClientPlatformSettings.Timeout`·`HttpClientProfileFactory`·testFixtures `ClientProfiles`에서 `streamingIdle` 제거, `docs/httpclient/env-fields.yaml`·`configuration-reference.md` 행 삭제. `HttpClientPlatformEnvManifestTest.theManifestAndTheSettingsTreeAgree` PASS(manifest와 record 트리 일치). 이제 `APP_HTTPCLIENT_CLIENTS_0_TIMEOUT_STREAMING_IDLE`을 설정하면 strict binder가 이름으로 거부 |
| P22-new-2 | 완료 | `b509991a`: `infra/redis-sdk/acl/application.acl`·`application-advanced.acl`·`all-accounts.acl`의 application 두 계정에서 `+@pubsub`·`&prod:*` 삭제(`resetchannels`만 남음). admin-readonly의 `+pubsub|channels` 등 introspection은 그대로(읽기 전용 관측). topology lane 결과는 아래 검증 절 |
| S3-lead-2 | 완료 | `68cda1d3`: env 오타 검사를 `StrictSettingsBinder`(support)로 일반화하고 httpclient `HttpClientEnvironmentKeys`(리플렉션 역산) 삭제. system env는 unbound 검사에서 빼고(Spring 열거형 `app.outbox.poll.interval`이 하이픈 필드를 오판 — 모든 트리가 `app.*`가 되면서 모든 capability의 문제가 됨), prefix의 env 형태로 시작하는 변수가 (a) 이번 바인딩이 바인딩한 속성, (b) 다른 source가 선언한 속성, (c) placeholder가 읽는 변수 중 하나의 env 형태가 아니면 이름으로 거부. `StrictSettingsBinderTest`: `aCorrectlySpelledHyphenatedEnvironmentVariableBinds`(FAIL: `BindException: Failed to bind properties under 'sample.shared'`)→PASS, `aMisspelledEnvironmentVariableIsRefusedByName`(FAIL: `Expecting actual throwable to be an instance of CapabilityStartupValidationException but was BindException`)→PASS, `aVariableASettingReadsThroughAPlaceholderIsNotATypo`(FAIL: BindException)→PASS, `aSubtreeAnotherModuleOwnsIsLeftToThatModule`(FAIL: BindException)→PASS, `aHyphenatedListFromTheEnvironmentBindsWhenForeignKeysAreExcluded`(P20-new-2 회귀). httpclient `HttpClientPlatformEnvManifestTest.aMisspelledEnvironmentVariableIsRefused` PASS(공용 binder로). manifest 도출(`fieldToEnvTemplate`)은 테스트로 옮김. `src/adapter/outbound/support/src/main/java/dev/caskeleton/adapter/outbound/support/StrictSettingsBinder.java` |
| X2-P1-3c | 완료 | 위 "strict 바인딩 현황" 표. 각 모듈 재현 테스트 FAIL(`Expecting: <Started application …> to have failed`)→PASS: `DistributedLockPersistenceConfigTest.aMisspelledLockKeyFailsStartup`, `JpaStartupAutoConfigurationTest.aMisspelledSafetyKeyFailsStartup`, `PersistenceVendorSelectionTest.aMisspelledPersistenceKeyFailsStartup`(`0f8e9053`), `RedisCapabilityCompositionTest.aMisspelledCapabilityKeyFailsStartup`(+`theIdempotencyLifecycleKeysAreLeftToTheirOwner`, `34fe03d1`), `NotificationPlatformCoreConfigurationTest.aMisspelledPlatformKeyFailsStartup`(`6ef946a0`), `MessagingPlatformBridgeAutoConfigurationTest.aMisspelledTopLevelMessagingKeyFailsStartup`(+`thePlatformsOwnSectionsAreLeftToThePlatform`, `2f19accc`), `ServiceRuntimeRegistrationTest.aMisspelledRuntimeKeyFailsStartup`(`976fcf0b`), `ObjectStorageCapabilityConfigTest.aMisspelledObjectStorageKeyFailsStartup`(`28fdbfb1`), `WebSocketSettingsTest.aMisspelledWebSocketKeyFailsStartup`·`GraphQlDeploymentModeContractTest.aMisspelledGraphQlKeyFailsStartup`·`MongoPersistenceConfigTest.aMisspelledPlatformKeyFailsStartup`(`8ff4371d`). graphql·websocket·mongo는 support edge가 없어(inbound→outbound 금지, mongo closure) Spring 고유 `ignoreUnknownFields=false` — yml/properties 오타는 거부, env 오타는 Spring이 system env를 검사에서 빼므로 거부하지 않음(차이를 README 대신 여기 기록). objectstorage는 registry에 support edge 추가(`28fdbfb1`, 커밋 메시지에 이유). `StrictSettingsBinder`는 `bindOrCreate`로(키가 없으면 기본값) |
| P22-new-1 | 완료 | `842e2402`: `JpaStartupAutoConfiguration#jpaProductionSecrets`가 `RequiredProductionSecrets`로 `APP_DATASOURCE_PASSWORD`를 기여(모듈이 켜져 있을 때만). `SecretSourceValidator`는 `app.persistence-jpa.enabled`를 더 읽지 않고 `REQUIRED_PROD_SECRETS = [APP_PRIVACY_PSEUDONYMIZATION_SALT]`. 레지스트리 `secrets-classification.yaml`의 `APP_DATASOURCE_PASSWORD`에 `required_when: app.persistence-jpa.enabled=true`(`SecretsClassificationRegistryTest` 1:1 유지). `JpaStartupAutoConfigurationTest.theModuleContributesTheDatabasePasswordAsAProductionSecret` FAIL(`Expecting … to have a single bean of type RequiredProductionSecrets`)→PASS, `aSwitchedOffModuleOwesNoDatabasePassword` PASS, bootstrap `SecretSourceValidatorTest.theGuardAloneDoesNotRequireADatabasePassword`(스위치 true + 빈 비밀번호 + 기여 없음 → 기동) PASS |
## 제거된 코드
| 대상 | 경로 | 마지막으로 존재한 커밋 | 이유 |
|---|---|---|---|
| `HttpClientEnvironmentKeys`(설정 record 트리를 리플렉션으로 순회해 env 이름 역산·거부) | `src/adapter/outbound/httpclient/src/main/java/dev/caskeleton/adapter/outbound/httpclient/autoconfigure/HttpClientEnvironmentKeys.java` | `0ebb77b5` | S3-lead-2: env 오타 거부를 `StrictSettingsBinder`가 모든 capability에 대해 한다. manifest 도출은 `HttpClientPlatformEnvManifestTest`로 |
| `StrictSettingsBinder.bind(env, prefix, type, Predicate<ConfigurationPropertySource>)` overload | `src/adapter/outbound/support/.../StrictSettingsBinder.java` | `0ebb77b5` | httpclient 전용 우회였다. system env 제외가 기본 동작이 됨 |
| 설정 키 `app.httpclient.clients[N].timeout.streaming-idle`(`APP_HTTPCLIENT_CLIENTS_N_TIMEOUT_STREAMING_IDLE`)과 `TimeoutSettings.streamingIdle`·`HttpClientPlatformSettings.Timeout.streamingIdle` | `src/adapter/outbound/httpclient/.../profile/TimeoutSettings.java`, `docs/httpclient/env-fields.yaml` | `0ebb77b5` | P11-new-1: 바인딩만 되고 읽는 곳 없음(capability 규칙 3) |
| yml 키 `ca-skeleton.rate-limit.{limit,window,algorithm}` | `src/app-bootstrap/src/test/resources/application-test.yml`, `src/sample-portfolio/src/systemTest/resources/application-test.yml` | `0ebb77b5` | 읽는 코드 없음. `app.rate-limit` 아래로 옮기면 web `EdgeRateLimitTransportSettings` 트리에 섞인다 |
| `SecretSourceValidator`의 `RELATIONAL_PERSISTENCE_PROPERTY`·`isRelationalPersistenceOn()`(모듈 스위치 해석) | `src/app-bootstrap/src/main/java/dev/caskeleton/bootstrap/runtime/SecretSourceValidator.java` | `0ebb77b5` | P22-new-1: persistence-jpa가 기여 |
| 설정 키 `app.notification.platform.smtp.max-concurrency`(이름 변경 → `dispatch-concurrency`, env `APP_NOTIFICATION_PLATFORM_SMTP_DISPATCH_CONCURRENCY` 그대로) | `NotificationSmtpSettings.java`, `notification.yml` | `0ebb77b5` | P20-new-1 |
| Redis ACL `+@pubsub`·`&prod:*`(application 계정 2개) | `infra/redis-sdk/acl/*.acl` | `0ebb77b5` | P22-new-2 |
## 새로 발견한 결함
- **P20-new-1** (고침, `6ef946a0`): `app.*` 통일로 `app.notification.platform.smtp.max-concurrency`(dispatch pool)의 relaxed env 형태가 `APP_NOTIFICATION_PLATFORM_SMTP_MAX_CONCURRENCY` — SMTP provider profile(`providers.smtp.max-concurrency`)의 placeholder 변수 — 와 같아져 한 변수가 두 속성을 설정하게 될 뻔했다(env가 yml보다 우선이라 `…_DISPATCH_CONCURRENCY`가 조용히 무시됨). 속성을 env 이름에 맞춰 `dispatch-concurrency`로. 전체 shipped yml·registry를 대상으로 "placeholder로 한 속성을 채우는 env가 relaxed 바인딩으로 다른 속성도 채우는가"를 스크립트로 확인해 남은 충돌 0건(sample의 죽은 `app.cache.redis.*` 한 건 제외 — 아래).
- **P20-new-2** (고침, `68cda1d3`): `StrictSettingsBinder.bindExcept`가 system environment까지 filter된 view로 감쌌다. filter된 view는 Spring의 env 이름 매핑을 잃어, 하이픈이 든 list/map(`APP_…_ALLOWED_HOSTS_0`)이 env에서 오면 오류 없이 빈 값으로 바인딩됐다(composition:idempotency가 이 경로). system env는 filter하지 않게. `StrictSettingsBinderTest.aHyphenatedListFromTheEnvironmentBindsWhenForeignKeysAreExcluded`.
- **P20-new-3** (남음): sample `application.yml`의 `app.cache.redis.*`(22개)·`app.outbound.http.*`(15개)·`app.messaging.kafka.brokers`·`app.notification.{email,slack}.provider`는 sample 조립에서 아무것도 바인딩하지 않는 죽은 키다(sample은 cache-redis·httpclient·notification을 싣지 않는다). sample 패키지는 P30 범위라 손대지 않음.
- **P20-new-4** (남음): `src/README.md` "Outbound HTTP client" 절의 yaml 예시(`app.capabilities.http-client.expected-state`·`app.providers.http-client`)는 존재하지 않는 트리를 설명한다(prefix만 이번에 바뀜). Wave 5 문서 정리 대상.
- **P20-new-5** (남음): persistence-jpa는 `app.persistence-jpa`(스위치)·`app.persistence`(vendor)·`app.jpa-platform`·`app.lock`·`app.outbox`·`app.realtime.replay`·`app.runtime.migration-on-startup`에 걸쳐 있다 — capability 규칙 4("prefix 하나")를 여전히 어긴다. `app.realtime.replay.retention`은 `@Value`라 strict 바인딩 밖. 키 재배치는 env·운영 문서 전체를 건드리는 별도 결정.
## 검증
- `./gradlew check --continue --max-workers=3` — BUILD SUCCESSFUL(499 tasks)
- `architectureCheck :app-bootstrap:verifyEnvKeys configContractCheck verifyDependencyLocks verifyPublicPathSnapshot` — 첫 실행에서 architectureTest 2건 실패(D3: `DistributedLockPersistenceConfig.lockSettings`에 스위치 조건 없음, B7: bridge 설정 bean이 public) → `e3e02052`로 고친 뒤 BUILD SUCCESSFUL. verifyEnvKeys: OK — 346 env keys
- `mainCheck :app-bootstrap:integrationTest :app-bootstrap:systemTest :sample-portfolio:integrationTest :sample-portfolio:systemTest` — BUILD SUCCESSFUL
- Redis topology lane(standalone, compose project `p20-redis`, 포트 16379, 끝나고 `down -v`): redis-runtime `LiveRedisCompositionTest` 7, redis-sdk `LiveRedisGuardrailTest` 6·`RedisTopologyContractTest` 8, cache-redis `LiveRedisSemanticScriptAclProbeTest` 2·`LiveRedisDistributedLeaseContractTest` 7·`LiveRedisSemanticPortsTest` 10 — 40개 전부 PASS. 추가로 `ACL DRYRUN ca-skeleton-application publish prod:x` → "no permissions", `get prod:x` → OK
- `integrationCheck --continue` — BUILD SUCCESSFUL(12m 18s; 입력이 없는 lane은 설계대로 빠짐)
- `-p optional-platforms ci` — 첫 실행 spotless 1건(`GrpcAdvancedModuleGuardTest`, 키 이름이 짧아져 줄바꿈 변화) → `13e624bd` 후 BUILD SUCCESSFUL
- **로컬 기동**(`:app-bootstrap:qualificationBootJar`, `SPRING_PROFILES_ACTIVE=local`, `src/.env.example` 값): 모든 capability가 켜진 jar라 인프라 없이 차례로 막히는 곳을 하나씩 껐다 — mongo(active-profile 필요), JPA(PostgreSQL 없음), security issuer 필수, gRPC insecure-local(이제 `APP_GRPC_SERVER_ALLOW_INSECURE_LOCAL`이 relaxed 바인딩으로 `app.grpc.server.allow-insecure-local`에 닿는 것 확인), notification(JPA 필요), Redis 자격 증명, fileserver fingerprint key, httpclient client 없음, Kafka broker 주소, object-storage 바인딩 — 모두 설정 바인딩 오류가 아닌 기존의 이름 붙은 기동 검사. 이 과정에서 결함 하나 발견·수정(P20-new-6). `APP_HTTPCLIENT_CLIENTS_0_TIMEUOT_TOTAL_CALL`을 넣으면 `unrecognised environment variables under app.httpclient: [APP_HTTPCLIENT_CLIENTS_0_TIMEUOT_TOTAL_CALL]`로 거부, 올바른 이름이면 통과. 마지막 조합에서 `Started CaSkeletonApplication in 10.7 seconds`(60초 timeout으로 종료)
- 수정 뒤 `:adapter:inbound:graphql:check` PASS
## 새로 발견한 결함 (추가)
- **P20-new-6** (고침, `12c61a99`): graphql을 `ignoreUnknownFields=false`로 바꾸자, 필수·무기본값인 `app.graphql.deployment-mode`가 shipped `graphql.yml`의 `${APP_GRAPHQL_DEPLOYMENT_MODE:}`로 빈 값이 되어 "left unbound" 바인딩 오류로 먼저 실패 — `GraphQlDeploymentMode`의 이름 붙은 거절("app.graphql.deployment-mode is required … set APP_GRAPHQL_DEPLOYMENT_MODE")을 가렸다(qualification jar 로컬 기동에서 발견). `GraphQlActivationEnvironmentPostProcessor`가 해석값이 빈 그 항목을 `graphql-defaults` source에서 뺀다. `GraphQlCapabilityActivationTest.anUnsetModeIsRefusedByNameRatherThanAsAnUnboundKey` FAIL(`Expecting actual … UnsatisfiedDependencyException … graphQlPlatformConfigurationCheck` 에 이름 붙은 문구 없음)→PASS. 같은 성질(빈 placeholder + 비-String 필드 + strict)의 다른 곳은 shipped yml 전수 확인에서 없음(나머지 빈 기본값은 모두 String/list).
- 마지막 커밋(`12c61a99`) 뒤 `check architectureCheck :app-bootstrap:verifyEnvKeys verifyDependencyLocks --continue` — BUILD SUCCESSFUL. `git status` 깨끗
통합 시 주의: 파일 전반(256개)이 키 이름을 바꾸므로 P30과 겹치는 파일(조건 annotation, sample yml 등)은 P30 쪽 새 코드에 `ca-skeleton.`/`backend.graphql`이 남지 않았는지 `git grep -nE '(^|[^/\w.-])ca-skeleton\.[a-z]|backend\.(graphql|web\.(mvc|advanced))' -- src`로 확인할 것(남아야 하는 것: 메시징 hash domain 4개, gRPC Context key 1개). `ConfigurationNamespaceContractTest.everyLeafPrefixSitsUnderTheApplicationRoot`가 설정 prefix 쪽은 잡는다.
@@ -71,6 +71,7 @@ Wave 3 라운드 1 결과로 받은 4건(2026-09-28, desktop):
| 33 | 분리된 persistence leaf의 caller 트랜잭션 guard | `persistence-notification`·`persistence-fileserver` → `persistence-jpa` edge 추가, `CallerTransaction` 하나를 쓴다 | X1-#4, X1-지도-caller 트랜잭션 guard |
| 34 | notification·fileserver id 포트를 domain `IdFactory`로 | 하지 않음 — 두 application leaf에 domain-core edge를 주지 않는다 → 보류 | P16-new-5 |
| 35 | X2-P2-6b 나머지(principal tenant, GraphQL Spring Security resolver, 가명화 키) | 전부 하지 않음 — 소비자 없는 필드는 만들지 않고, adopter resolver bean 방식 유지, 키 분리는 의도(문서화) → 보류 | J-12 / X2-P2-6b |
| 36 | persistence-jpa 설정 prefix 여러 개 | 기능별 분리 인정 — JPA 플랫폼 자체(스위치·vendor·jpa-platform·migration-on-startup)는 `app.persistence-jpa.*` 하나로, 여러 저장소가 공유하는 outbox·lock·replay는 기능 트리(`app.outbox`·`app.lock`·`app.realtime.replay`)로 두고 규칙 4에 이 예외를 적는다. `APP_*` env 키 유지 | P20-new-5 |
기존 원칙에서 바로 따라 나와 에이전트가 정한 것(소유자가 뒤집을 수 있음):
@@ -735,7 +736,7 @@ Wave 3 라운드 1 결과로 받은 4건(2026-09-28, desktop):
| X2-P1-2b | Required | httpclient `profile/RuntimeEnvironment`가 정본과 이름이 같음(Nit) | 남음 | `adapter/outbound/httpclient/.../profile/RuntimeEnvironment.java` (httpclient 안 참조 파일 4개 + 정본 import 1개) | W5 | `HttpClientHardening` 등으로 개명, 참조 4파일 수정 |
| X2-P1-3a | Required(문서) | capability settings 규칙(strict 바인딩 + compact constructor)을 root 문서에 | 완료 | `AGENTS.md:216` capability 완료 기준 4, 커밋 `6d27356e` | - | - |
| X2-P1-3b | Required | gRPC adapter strict prefix `ca-skeleton.grpc`가 벤더 `ca-skeleton.grpc.platform`과 겹침 | 완료 | Wave 1–2 P5 (fcddab7e·Wave 2 커밋): `GrpcSafeActivationTest.theVendorPlatformNamespaceBesideTheAdapterDoesNotFailItsStrictBinding` FAIL(`UnboundConfigurationPropertiesException: The elements [ca-skeleton.grpc.platform.enabled] were left unbound`)→PASS. 어댑터 prefix `ca-skeleton.grpc` → `ca-skeleton.grpc.server` `src/adapter/inbound/grpc/src/main/java/dev/caskeleton/adapter/inbound/grpc/GrpcServerSettings.java:17`, `GrpcServerConfig.java`. 어댑터 키는 env-keys registry에 행이 없었다(추가·수정 없음). 갱신: `src/adapter/inbound/grpc/{CLAUDE,README}.md`, `src/README.md:269`(env `CA_SKELETON_GRPC_SERVER_ALLOW_INSECURE_LOCAL`). **오케스트레이터 할 일:** `docs/capability-status.md:42`의 `ca-skeleton.grpc.enabled`·`ca-skeleton.grpc.allow-insecure-local`을 `.server.` 경로로. Wave 3-B(`app.*`)에서도 `app.grpc.server`처럼 벤더 prefix와 형제로 둬야 같은 충돌이 다시 안 생긴다. 커밋 `a498ad84` | - | adapter prefix를 `ca-skeleton.grpc.server` 등으로 분리(설정 키·env key·README 갱신) |
| X2-P1-3c | Required | 오타 키 거부가 일부 capability에만 적용 | 남음 | strict 바인딩 4곳(fileserver·httpclient `StrictSettingsBinder`, `GrpcServerSettings.java:17`, `GrpcPlatformProperties.java:19`) / main `@ConfigurationProperties` 54파일(`git grep -l "@ConfigurationProperties" -- 'src/**/src/main/java/**'`) | W3 | `AGENTS.md:216` 기준대로 capability 모듈(cache-redis, persistence-jpa, persistence-mongo, messaging, notification, objectstorage, graphql, websocket) settings를 `StrictSettingsBinder`로 바인딩 |
| X2-P1-3c | Required | 오타 키 거부가 일부 capability에만 적용 | 완료 | Wave 3 W3-P20 (Wave 3 커밋): 위 "strict 바인딩 현황" 표. 각 모듈 재현 테스트 FAIL(`Expecting: <Started application …> to have failed`)→PASS: `DistributedLockPersistenceConfigTest.aMisspelledLockKeyFailsStartup`, `JpaStartupAutoConfigurationTest.aMisspelledSafetyKeyFailsStartup`, `PersistenceVendorSelectionTest.aMisspelledPersistenceKeyFailsStartup`(`0f8e9053`), `RedisCapabilityCompositionTest.aMisspelledCapabilityKeyFailsStartup`(+`theIdempotencyLifecycleKeysAreLeftToTheirOwner`, `34fe03d1`), `NotificationPlatformCoreConfigurationTest.aMisspelledPlatformKeyFailsStartup`(`6ef946a0`), `MessagingPlatformBridgeAutoConfigurationTest.aMisspelledTopLevelMessagingKeyFailsStartup`(+`thePlatformsOwnSectionsAreLeftToThePlatform`, `2f19accc`), `ServiceRuntimeRegistrationTest.aMisspelledRuntimeKeyFailsStartup`(`976fcf0b`), `ObjectStorageCapabilityConfigTest.aMisspelledObjectStorageKeyFailsStartup`(`28fdbfb1`), `WebSocketSettingsTest.aMisspelledWebSocketKeyFailsStartup`·`GraphQlDeploymentModeContractTest.aMisspelledGraphQlKeyFailsStartup`·`MongoPersistenceConfigTest.aMisspelledPlatformKeyFailsStartup`(`8ff4371d`). graphql·websocket·mongo는 support edge가 없어(inbound→outbound 금지, mongo closure) Spring 고유 `ignoreUnknownFields=false` — yml/properties 오타는 거부, env 오타는 Spring이 system env를 검사에서 빼므로 거부하지 않음(차이를 README 대신 여기 기록). objectstorage는 registry에 support edge 추가(`28fdbfb1`, 커밋 메시지에 이유). `StrictSettingsBinder`는 `bindOrCreate`로(키가 없으면 기본값) | - | `AGENTS.md:216` 기준대로 capability 모듈(cache-redis, persistence-jpa, persistence-mongo, messaging, notification, objectstorage, graphql, websocket) settings를 `StrictSettingsBinder`로 바인딩 |
| X2-P1-4a | Consider | web `cursor/CursorCodec`(dev key 하드코딩) 삭제 | 완료 | `docs/capability-status.md:94`, `HmacWebCursorCodec`도 삭제(`:97`), 커밋 `6d27356e` | - | - |
| X2-P1-4b | Consider | 서명 계층 하나(inbound에서 서명, persistence는 typed keyset), 남으면 공용 서명 헬퍼 | 완료 | Wave 3 W3-P19 (Wave 3 라운드 3 커밋): 커밋 `35e1df8a`. `shared-contract/.../digest/HmacSha256.java`(sign/verify, 길이 먼저 확인 후 constant-time), `HmacSha256Test` 5건. 사용: graphql `HmacGraphQlCursorCodec`·`GraphQlIdentityFingerprinter`, persistence-jpa `SignedJsonCursorCodec`, notification HMAC 3곳, sample digester. 서명 codec은 유지(2026-09-25 결정). mongo `MongoKeysetCursorCodec`은 shared-contract edge 없어 유지, redis·identifier·fileserver는 동시 묶음/edge 없음으로 유지 | - | 결정(2026-09-25): Wave 3 라이브러리 유지 결정을 따름 — 서명 codec 유지, shared-contract에 sign/verify 헬퍼 하나로 중복 제거 |
| X2-P1-5 | Consider | SHA-256 보일러플레이트 → shared-contract `Sha256` | 완료 | Wave 3 W3-P19 (Wave 3 라운드 3 커밋): 커밋 `bca2c682`. `shared-contract/.../digest/Sha256.java`(of/ofUtf8/newMessageDigest/hex), `Sha256Test` 5건. `MessageDigest.getInstance` main 36곳 → 10곳(mongo 3, application-objectstorage 2 — shared-contract edge 없음; vendored messaging 2·gRPC 2; cache-redis SHA-1 1) | - | shared-contract `Sha256.hex(String)`·`digest(byte[])` 추가, edge 있는 25파일 교체(objectstorage 12, sample 3, graphql 2, cache-redis 2, fileserver 2, notification 2, persistence-jpa 2); mongo 3·application-objectstorage 2(edge 없음)와 벤더 5는 유지 |
@@ -860,9 +861,9 @@ Wave 3 라운드 1 결과로 받은 4건(2026-09-28, desktop):
| S3-#13b | Nit | draining·retry 소진을 `HttpRateLimitRejectedException`으로 던짐(텔레메트리 카테고리 오기) | 완료 | Wave 1–2 P11 (fcddab7e·Wave 2 커밋): `DefaultGenericHttpGatewayTest.anExhaustedRetryBudgetIsReportedAsItsOwnFailureNotARateLimit`·`aRetryRefusedByADrainingRuntimeIsReportedAsItsOwnFailureNotARateLimit` FAIL(`expected HttpRetryBudgetExhaustedException/HttpRuntimeDrainingException but was HttpRateLimitRejectedException`)→PASS. `HC/restclient/DefaultGenericHttpGateway.java:183,281`, 새 타입 `HC/api/error/HttpRetryBudgetExhaustedException.java`·`HttpRuntimeDrainingException.java`, `docs/httpclient/operations.md` 예외 표 갱신. 두 경로는 재시도 판정과 다음 attempt 사이의 경합에서만 도달하므로 테스트는 `TestGateways.forProfile(profile, transport, eligibility)`로 판정기를 주입해 재현 | - | draining·retry 소진 전용 예외(또는 카테고리)로 분리 |
| S3-#13c | Nit | Redis health indicator 둘이 DOWN 값만 다름 | 남음 | `redis-runtime/.../sdk/config/RedisSdkAutoConfiguration.java:257`(DEGRADED)·`:285`(DOWN) | W5 | 상태 값만 인자로 받는 indicator 하나로 |
| S3-lead-1 | 권고 | `RedisSecretSource` + `System.getenv` 기본값 vs `SecretSourcePort` (= X2 P1-1) | 완료 | 문제 아님: redis-runtime은 의존 edge가 없어 자체 SPI(`RedisSdkAutoConfiguration.java:168-171,334`)를 두고, cache-redis가 `SecretSourcePort`로 bridge(`cache-redis/.../capability/RedisCapabilityConfig.java:66`). X2 P1-1이 이 구조를 모범 사례로 판정 | - | - |
| S3-lead-2 | 권고 | `HttpClientEnvironmentKeys`(리플렉션 env 이름 역산) vs `StrictSettingsBinder` (= X2 P1-3) | 남음 | `httpclient/.../autoconfigure/HttpClientEnvironmentKeys.java`(179줄, `RecordComponent` 순회), `HttpClientPlatformSettingsBinder.java:27`만 사용. 다른 capability는 env 오타 검사 없음 | W3 | env 이름 역산 검사를 support `StrictSettingsBinder` 옵션으로 옮겨 모든 capability가 같은 방식으로 오타 env를 거부 |
| S3-lead-2 | 권고 | `HttpClientEnvironmentKeys`(리플렉션 env 이름 역산) vs `StrictSettingsBinder` (= X2 P1-3) | 완료 | Wave 3 W3-P20 (Wave 3 커밋): `68cda1d3`: env 오타 검사를 `StrictSettingsBinder`(support)로 일반화하고 httpclient `HttpClientEnvironmentKeys`(리플렉션 역산) 삭제. system env는 unbound 검사에서 빼고(Spring 열거형 `app.outbox.poll.interval`이 하이픈 필드를 오판 — 모든 트리가 `app.*`가 되면서 모든 capability의 문제가 됨), prefix의 env 형태로 시작하는 변수가 (a) 이번 바인딩이 바인딩한 속성, (b) 다른 source가 선언한 속성, (c) placeholder가 읽는 변수 중 하나의 env 형태가 아니면 이름으로 거부. `StrictSettingsBinderTest`: `aCorrectlySpelledHyphenatedEnvironmentVariableBinds`(FAIL: `BindException: Failed to bind properties under 'sample.shared'`)→PASS, `aMisspelledEnvironmentVariableIsRefusedByName`(FAIL: `Expecting actual throwable to be an instance of CapabilityStartupValidationException but was BindException`)→PASS, `aVariableASettingReadsThroughAPlaceholderIsNotATypo`(FAIL: BindException)→PASS, `aSubtreeAnotherModuleOwnsIsLeftToThatModule`(FAIL: BindException)→PASS, `aHyphenatedListFromTheEnvironmentBindsWhenForeignKeysAreExcluded`(P20-new-2 회귀). httpclient `HttpClientPlatformEnvManifestTest.aMisspelledEnvironmentVariableIsRefused` PASS(공용 binder로). manifest 도출(`fieldToEnvTemplate`)은 테스트로 옮김. `src/adapter/outbound/support/src/main/java/dev/caskeleton/adapter/outbound/support/StrictSettingsBinder.java` | - | env 이름 역산 검사를 support `StrictSettingsBinder` 옵션으로 옮겨 모든 capability가 같은 방식으로 오타 env를 거부 |
| S3-lead-3 | 권고 | TLS material 로딩이 Redis와 httpclient에 따로 | 완료 | 문제 아님: Redis는 Lettuce `SslOptions`에 PEM 위치만 넘김(`RedisTopologyClientFactory.java:327-362`), httpclient는 JSSE `KeyStore`·`TrustManagerFactory` 구성(`security/TlsMaterialProvider.java`). 클라이언트 API가 달라 공유할 코드가 없고 redis-runtime은 의존 edge 0 | - | - |
| S3-lead-4 | 권고 | `ca-skeleton.capabilities.*` selector 문자열이 여러 leaf에 흩어짐 | 남음 | main 13파일(sample, app-bootstrap 3, web 1, persistence-jpa 3, redis-runtime `RedisCorrectnessRoles.java`, cache-redis 4), `...idempotency.provider` 30회 | W3 | 역할별 selector 키를 한 곳(멱등성은 S1-lead-3의 composition 모듈, Redis는 `RedisCorrectnessRoles`)에 상수로 두고 나머지는 그 상수/조건 클래스를 사용 |
| S3-lead-4 | 권고 | `ca-skeleton.capabilities.*` selector 문자열이 여러 leaf에 흩어짐 | 완료 | Wave 3 W3-P20 (Wave 3 커밋): `711a3968`: `src/adapter/outbound/support/src/main/java/dev/caskeleton/adapter/outbound/support/CapabilitySelectors.java`(ROOT·CACHE_DEFAULT_BINDING·IDEMPOTENCY_PROVIDER·RATE_LIMIT_PROVIDER·LEASE_PROVIDER). 사용: cache-redis `RedisCacheConfig`·`RedisCorrectnessRoleSelectors`·`RedisIdempotencyConfig`·`RedisLeaseConfig`·`RedisRateLimitConfig`·`RedisCapabilitySettings.bind`, persistence-jpa `PostgreSqlIdempotencyStoreConfig`·`IdempotencyReaper`·`IdempotencySchemaStream`, composition:idempotency `IdempotencyAutoConfiguration`. 현황표 방향(멱등성은 composition 모듈에)과 다르게 support에 둔 이유: persistence-jpa·cache-redis는 composition:idempotency에 의존할 수 없고(registry), 셋 다 support에 이미 의존한다. main의 `app.capabilities.*.provider` 문자열 리터럴 0건(`git grep '"app.capabilities' -- '*/src/main/*'`는 메시지·주석뿐) | - | 역할별 selector 키를 한 곳(멱등성은 S1-lead-3의 composition 모듈, Redis는 `RedisCorrectnessRoles`)에 상수로 두고 나머지는 그 상수/조건 클래스를 사용 |
| S3-lead-5 | 권고 | `secret://<source>/<name>` 파싱이 다른 leaf에도 | 완료 | Wave 3 W3-P26 (Wave 3 라운드 2 커밋): `application-core` `SecretReference`(record source/name, `parse`·`isReference`, `application-core/.../outbound/SecretReference.java:47`)를 `SecretSourcePort` 옆에 추가. httpclient(`HttpClientAuthenticationAutoConfiguration.java:60-64`)와 cache-redis(`RedisKeys.java:83`)가 사용 — 둘 다 기존 edge(application-core) 안, registry 변경 없음. mongo·redis-runtime은 지시대로 유지. 테스트 `SecretReferenceTest` 6개(새 타입이라 FAIL은 컴파일 실패). cache-redis는 이름이 빈 `secret://src/`를 빈 키로 port에 넘기던 것이 이제 거부됨(`referenceWhoseNameIsBlankIsRejected`) | - | application-core `SecretSourcePort` 옆에 `SecretReference.parse` 하나, httpclient·cache-redis가 사용(mongo·redis-runtime은 edge 없어 유지) |
**조사 메모**
@@ -992,7 +993,7 @@ Wave 3 라운드 1 결과로 받은 4건(2026-09-28, desktop):
| S6-#11c | Consider | sample 테스트가 `TransactionPort` fake를 각자 작성 (= X1-#3 일부) | 완료 | application-core testFixtures `DirectTransactionPort`(`src/application-core/src/testFixtures/java/dev/caskeleton/application/transaction/testkit/DirectTransactionPort.java`), sample 테스트의 인라인 fake 0건, `6d27356e` | - | - |
| S6-#12 | Consider | JPA 정책 클래스 겸용, 주입처 없는 bean, 테스트만 쓰는 플래그 | 완료 | `JpaPlatformAutoConfiguration` 삭제·`JpaStartupAutoConfiguration`으로 합침(`CS:238`), `app.jpa-platform.addons-enabled` 삭제(`CS:193`), `jpaDataSourceProfileValidator`가 `jpaResolvedDataSourceCheck`에 주입됨(`JS/JpaStartupAutoConfiguration.java:100,109-110`), `6d27356e` | - | - |
| S6-#13 | Consider | 역할 없는 startup 예외 subclass 4개 | 완료 | Wave 3 W3-P25 (Wave 3 라운드 2 커밋): 4개 subclass 삭제, `StartupFailureException`을 final + package-private 생성자로(`support/.../startup/StartupFailureException.java:18,25`), 생성은 `StartupFailures`만. 테스트 23개 파일의 `isInstanceOf(XxxException.class)`를 `isInstanceOfSatisfying(StartupFailureException.class, e -> errorCode 단언)`으로. `StartupFailureExceptionTest`는 코드→exit 파라미터 테스트. support·persistence-jpa·cache-redis·app-bootstrap `test`/`architectureTest` PASS | - | `StartupFailureException` 하나 + 테스트는 `errorCode()` 단언; 생성은 `StartupFailures` factory만 |
| S6-#14 | Consider | 스위치 prefix 3종, "capability" 뜻 여러 개 | 남음 | prefix: `ca-skeleton.*`(persistence-jpa, persistence-mongo, notification, outbox, grpc), `app.*`(messaging, redis, fileserver-platform, httpclient, object-storage, websocket), `backend.graphql`(`CS:30-45`). "capability": `@UseCaseCapability`, `capabilities.*` selector, `JpaCapability`, `MessagingCapabilities`. `MasterSwitch`·endpoint 미보고 부분은 삭제로 해당 없음 | W3 | 결정(2026-09-25): 설정 prefix를 `app.*`로 통일(`ca-skeleton.*`, `backend.graphql` → `app.*`), `APP_*` env 키 유지 |
| S6-#14 | Consider | 스위치 prefix 3종, "capability" 뜻 여러 개 | 완료 | Wave 3 W3-P20 (Wave 3 커밋): 결정 16: 설정 prefix를 `app.*` 하나로(위 표). `711a3968`·`fabb611e`. env 키 이름 변경 없음. `ConfigurationNamespaceContractTest`: 허용 root는 `app` 하나(`shippedConfigurationDeclaresOnlyRecordedRoots`), 모든 leaf의 `@ConfigurationProperties`·`String PREFIX` 상수가 `app.` 아래(`everyLeafPrefixSitsUnderTheApplicationRoot`, 새 테스트) — `src/app-bootstrap/src/test/java/dev/caskeleton/bootstrap/contract/ConfigurationNamespaceContractTest.java`. yml root 병합 4곳(cache-redis·notification module-defaults, sample `application.yml`·systemTest yml). "capability" 용어 정리는 이번 범위 밖(키 이름만) | - | 결정(2026-09-25): 설정 prefix를 `app.*`로 통일(`ca-skeleton.*`, `backend.graphql` → `app.*`), `APP_*` env 키 유지 |
| S6-#15 | Consider | bootstrap `:test`에 저장소 governance 테스트 | 남음 | `src/app-bootstrap/src/test/java/dev/caskeleton/bootstrap/contract/` 하위 45파일 7,302줄(`208077fd` 50파일), 이 중 34개가 저장소 파일·리소스를 직접 읽음 | W4 | `architectureTest`처럼 별도 strict lane으로 옮기고 읽는 파일을 `inputs`로 선언 |
| S6-#16a | Nit | 이력·리뷰 ID 주석 | 남음 | bootstrap main 이력/ID 줄 35개(105에서 감소, 주석 비율 19%), websocket 2개(해소 수준) | W5 | bootstrap 남은 35줄 삭제 |
| S6-#16b | Nit | `CapabilityDependencyEnvironmentValidator` 30줄 Javadoc | 해당 없음(코드 삭제) | `CS:249` | - | - |
@@ -1166,7 +1167,7 @@ Wave 1–2 구현 묶음(`P1`–`P13`)이 작업 중 찾은 결함과 통합 검
| P10-new-1 | 발견 | `src/.env.example:25`가 `APP_SERVER_FORWARD_HEADERS_STRATEGY=framework`를 권했다. `framework`는 Spring `ForwardedHeaderFilter`가 모든 peer의 `X-Forwarded-For`를 믿어 `getRemoteAddr()` 자체를 호출자가 고르게 하므로, rate-limit 모드와 상관없이 비인증 키를 우회할 수 있었다. `na… | 완료 | Wave 1–2 P10 (fcddab7e·Wave 2 커밋): 결과 파일 `docs/reviews/2026-09-25-review-followup-results/P10.md` | - | - |
| P10-new-2 | 발견 | `WebMvcRequestIdFilter`의 traceparent 정규식은 all-zero trace id 등을 받아들였고 `RequestLoggingFilter`의 `TraceParent.parse`는 거부해서, 같은 요청에서 두 필터가 다른 trace id를 가질 수 있었다(I-4b의 한 원인). 이제 `TraceParent.parse` 하나만 쓴다. | 완료 | Wave 1–2 P10 (fcddab7e·Wave 2 커밋): 결과 파일 `docs/reviews/2026-09-25-review-followup-results/P10.md` | - | - |
| P10-new-3 | 발견 | 옛 콜백 컨트롤러의 `@RequestBody byte[]`는 form POST(`application/x-www-form-urlencoded`, query string 없음)에서 Spring 7 `ServletServerHttpRequest.getBody()`가 요청 파라미터로 다시 만든 본문을 준다(`getBodyFromServletRequestParameters`, spring-web 7.0.9 bytec… | 완료 | Wave 1–2 P10 (fcddab7e·Wave 2 커밋): 결과 파일 `docs/reviews/2026-09-25-review-followup-results/P10.md` | - | - |
| P11-new-1 | 발견 | `app.httpclient.clients[N].timeout.streaming-idle`(`APP_HTTPCLIENT_CLIENTS_N_TIMEOUT_STREAMING_IDLE`)은 바인딩만 되고 main 코드 어디서도 읽지 않음(reactive 삭제 전에도 그랬음). `HC/profile/TimeoutSettings.java:25`, `docs/httpclient/configuration-reference… | 남음 | 결과 파일 `docs/reviews/2026-09-25-review-followup-results/P11.md` | W3 | 읽지 않는 `timeout.streaming-idle` 키·env 키 삭제 |
| P11-new-1 | 발견 | `app.httpclient.clients[N].timeout.streaming-idle`(`APP_HTTPCLIENT_CLIENTS_N_TIMEOUT_STREAMING_IDLE`)은 바인딩만 되고 main 코드 어디서도 읽지 않음(reactive 삭제 전에도 그랬음). `HC/profile/TimeoutSettings.java:25`, `docs/httpclient/configuration-reference… | 완료 | Wave 3 W3-P20 (Wave 3 커밋): `711a3968`: 읽지 않던 `clients[N].timeout.streaming-idle` 키 삭제 — `TimeoutSettings`·`HttpClientPlatformSettings.Timeout`·`HttpClientProfileFactory`·testFixtures `ClientProfiles`에서 `streamingIdle` 제거, `docs/httpclient/env-fields.yaml`·`configuration-reference.md` 행 삭제. `HttpClientPlatformEnvManifestTest.theManifestAndTheSettingsTreeAgree` PASS(manifest와 record 트리 일치). 이제 `APP_HTTPCLIENT_CLIENTS_0_TIMEOUT_STREAMING_IDLE`을 설정하면 strict binder가 이름으로 거부 | - | 읽지 않는 `timeout.streaming-idle` 키·env 키 삭제 |
| P11-new-2 | 발견 | `scripts/verify-httpclient-docs.py:22`가 없는 경로 `src/app-bootstrap/.../autoconfigure/httpclient/HttpClientStartupValidator.java`를 읽어 `FileNotFoundError`로 실패함. 경로를 `.../httpclient/autoconfigure`로 바꿔 돌리면 `rotation-drain-timeout`이 `doc… | 남음 | 결과 파일 `docs/reviews/2026-09-25-review-followup-results/P11.md` | W5 | `scripts/verify-httpclient-docs.py` 경로 수정, `rotation-drain-timeout` 문서 추가 |
| P11-new-3 | 발견 | `httpClientFailureInjectionTest`의 `FailureInjectionContractTest.disabledProxyFailsBeforeARequestCanBeSent`가 이 머신에서 실패(`expected: NOT_SENT but was: SENT_NO_RESPONSE`). Docker userland proxy(`docker-proxy` 동작 중)가 꺼진 Toxiproxy 리스너의 p… | 남음 | 결과 파일 `docs/reviews/2026-09-25-review-followup-results/P11.md` | W4 | Docker userland proxy에 의존하지 않는 방식으로 lane 수정(기준 커밋에서도 실패 — 기존 결함) |
| P11-new-4 | 발견 | Dynamic Target cross-origin hop은 `SensitiveHeaderStripper` 표준 이름(`authorization`·`cookie`·`x-api-key`·`api-key` 등)만 제거하므로, 호출자가 `HeaderPolicy`로 넘긴 관례 밖 이름의 자격증명 헤더(예: `X-Client-Key`)는 다음 origin으로 따라감 `HC/dynamic/DefaultDynamicTarg… | 완료 | Wave 2.5 W2-P14 (Wave 3 라운드 1 커밋): 결정 23. `DefaultDynamicTargetGatewayTest.crossOriginHopKeepsOnlySafeCallerHeaders` FAIL(`[(true, true), (true, true)]`)→PASS. `HC/dynamic/DefaultDynamicTargetGateway.java:164,172-185` cross-origin hop에서 호출자 헤더를 allowlist(`Accept`·`Accept-Language`·`Content-Type`·`Content-Language`)로만 남김. 자격증명 binding 헤더는 hop마다 새 target 기준으로 다시 붙음(기존) | - | 결정 23: cross-origin hop에서 호출자 헤더 전부 제거, 안전한 헤더만 유지 |
@@ -1215,8 +1216,8 @@ Wave 1–2 구현 묶음(`P1`–`P13`)이 작업 중 찾은 결함과 통합 검
| P21-new-1 | 발견 | handler가 `Retry`를 답하면 Kafka consumer가 시도 횟수를 세지 않아 partition이 영구히 멈췄다. | 완료 | Wave 3 P21: 예외와 같은 시도 예산·격리 사용 | - | - |
| P21-new-2 | 발견 | 다시 읽은 delivery가 항상 `deliveryAttempt=1`을 보고했다. | 완료 | Wave 3 P21 | - | - |
| P21-new-3 | 발견 | `src/messaging/CLAUDE.md`가 `MessagingPublicSurfaceContractTest`를 인용 — 테스트는 app-bootstrap에 있다(P21은 없다고 보고). 통합 때 이 테스트가 P21 변경(reactive facade 삭제, leaf 병합, starter-kafka 공개 consumer seam)을 반영하지 못해 실패 → 기대값 수정, starter-kafka에 `api kafka-clients` 선언. | 남음 | Wave 3 라운드 3 통합 | W5 | messaging CLAUDE.md의 테스트 위치 문구를 app-bootstrap 경로로 |
| P22-new-1 | 발견 | bootstrap `SecretSourceValidator`가 persistence-jpa의 모듈 스위치를 읽어 `APP_DATASOURCE_PASSWORD` 필요 여부를 판정(모듈 지식이 bootstrap에). | 남음 | Wave 3 P22 결과 파일 | W3 | Redis처럼 persistence-jpa가 secret 요구를 기여. P20 |
| P22-new-2 | 발견 | `infra/redis-sdk/acl/application.acl`·`all-accounts.acl`이 pub/sub 삭제 뒤에도 `+@pubsub`·`&prod:*`를 준다(최소 권한 위반, fixture). | 남음 | Wave 3 P22 결과 파일 | W3 | ACL fixture에서 제거 후 topology lane 재실행. P20 |
| P22-new-1 | 발견 | bootstrap `SecretSourceValidator`가 persistence-jpa의 모듈 스위치를 읽어 `APP_DATASOURCE_PASSWORD` 필요 여부를 판정(모듈 지식이 bootstrap에). | 완료 | Wave 3 W3-P20 (Wave 3 커밋): `842e2402`: `JpaStartupAutoConfiguration#jpaProductionSecrets`가 `RequiredProductionSecrets`로 `APP_DATASOURCE_PASSWORD`를 기여(모듈이 켜져 있을 때만). `SecretSourceValidator`는 `app.persistence-jpa.enabled`를 더 읽지 않고 `REQUIRED_PROD_SECRETS = [APP_PRIVACY_PSEUDONYMIZATION_SALT]`. 레지스트리 `secrets-classification.yaml`의 `APP_DATASOURCE_PASSWORD`에 `required_when: app.persistence-jpa.enabled=true`(`SecretsClassificationRegistryTest` 1:1 유지). `JpaStartupAutoConfigurationTest.theModuleContributesTheDatabasePasswordAsAProductionSecret` FAIL(`Expecting … to have a single bean of type RequiredProductionSecrets`)→PASS, `aSwitchedOffModuleOwesNoDatabasePassword` PASS, bootstrap `SecretSourceValidatorTest.theGuardAloneDoesNotRequireADatabasePassword`(스위치 true + 빈 비밀번호 + 기여 없음 → 기동) PASS | - | Redis처럼 persistence-jpa가 secret 요구를 기여. P20 |
| P22-new-2 | 발견 | `infra/redis-sdk/acl/application.acl`·`all-accounts.acl`이 pub/sub 삭제 뒤에도 `+@pubsub`·`&prod:*`를 준다(최소 권한 위반, fixture). | 완료 | Wave 3 W3-P20 (Wave 3 커밋): `b509991a`: `infra/redis-sdk/acl/application.acl`·`application-advanced.acl`·`all-accounts.acl`의 application 두 계정에서 `+@pubsub`·`&prod:*` 삭제(`resetchannels`만 남음). admin-readonly의 `+pubsub channels` 등 introspection은 그대로(읽기 전용 관측). topology lane 결과는 아래 검증 절 | - | ACL fixture에서 제거 후 topology lane 재실행. P20 |
| P22-new-3 | 발견 | redis-sdk 공개 API가 runtime 타입을 노출하면서 runtime을 `implementation`으로만 의존. `ca.spring-library`에 `api` 구성이 없어 한 줄로 못 고침. | 완료 | Wave 3 W3-P29 (Wave 3 라운드 3 커밋): `3967c9ef`. redis-sdk만 `java-library` + `api project(':adapter:outbound:redis-runtime')` / 통합 검증 통과(redis-sdk java-library + api redis-runtime) | - | 컨벤션에 `java-library` 적용 또는 공개 표면에서 runtime 타입 제거. P29 |
| ORCH-new-4 | 발견 | redis-sdk `RedisSdkModuleBoundaryTest`·`RedisSupportMatrixTest`가 옛 split package(`redis-runtime/.../redis/sdk`)를 훑음 — P22 worktree에서 통과, main 통합에서 `NoSuchFileException`(ORCH-new-2와 같은 유형: 저장소 파일을 읽지만 test inputs 미선언). | 남음 | 경로는 통합 때 수정(`redis-sdk` 하나) | W4 | 읽는 파일을 test `inputs`로 선언 |
| P19-new-1 | 발견 | notification `platform.api.TenantId`가 비어 있지 않음·100자 이하만 검사 — web·GraphQL이 거부하는 tenant(공백·줄바꿈·경로 문자)를 받는다. 강화하면 이미 저장된 행을 못 읽을 수 있다. | 완료 | Wave 3 W3-P30 (Wave 3 라운드 4 커밋): 커밋 `1fc862a8`. `application-notification/.../platform/api/TenantId.java`는 그대로(저장된 행을 읽는 타입 — 이전 규칙으로 쓴 행도 읽힘, `NotificationSubmissionServiceTest.aTenantStoredUnderTheOldRuleIsStillReadable`). 새 쓰기만 `dispatch/WritableTenant.require`(package-private)로 application-core `TenantId` 문법(`[A-Za-z0-9][A-Za-z0-9_.-]{0,63}`)을 요구, 어긋나면 `NotificationValidationException`(`VALIDATION_FAILED`, 다른 제출 검증과 같은 코드): `NotificationSubmissionService.prepare`(submit·schedule 공통 첫 단계), `AcceptNotificationApplicationUseCase.handle`(첫 쓰기인 연락처 등록 전). 테스트: `NotificationSubmissionServiceTest.aSubmissionForATenantOutsideTheTransportGrammarIsRefused`(공백·줄바꿈·`../`·`/`·선행 `-` 5건) FAIL(`Expecting actual not to be null`) → PASS, `AcceptNotificationApplicationUseCaseTest.anAcceptanceForATenantOutsideTheTransportGrammarRegistersNothing` FAIL → PASS. 기본 단일 tenant `"default"`(`NotificationPlatformCodecConfig.java:57`)는 문법 안. 검증: application-notification 230, adapter:outbound:notification·web-notification·composition notification `check` PASS | - | 읽기는 관대하게, 새 쓰기만 공용 `TenantId` 문법으로. P30 |
@@ -1227,3 +1228,9 @@ Wave 1–2 구현 묶음(`P1`–`P13`)이 작업 중 찾은 결함과 통합 검
| P30-new-1 | 발견 | fileserver `DefaultCleanupService.cleanStaging`이 `claimForCleanup`(JPA `@Modifying` update)을 트랜잭션 밖에서 불러 실제 store에서는 세션 있는 staging 정리가 매번 실패·재시도 후 포기됐다. | 완료 | Wave 3 P30 `4de7717d`: guard 추가 전 호출자를 트랜잭션 안으로 | - | - |
| P30-new-2 | 발견 | notification dispatch Blocked 분기가 recipient 전이 트랜잭션을 닫은 뒤 request 상태 roll-up을 트랜잭션 밖에서 실행. | 완료 | Wave 3 P30 `4de7717d` | - | - |
| P30-new-3 | 발견 | objectstorage local-dev provider의 create-only 게시가 이제 한 프로세스 안에서만 보장된다(JDK에 디렉터리 핸들 기준 create-only rename·link가 없어 검증된 부모에서 부재 확인+rename, JVM 단위 lock). 예전 hard-link 방식은 프로세스 간에도 보장했다. provider는 단일 프로세스·local-dev 전용으로 선언돼 있다. | 남음 | Wave 3 P30 결과 파일 | W5 | provider README에 단일 프로세스 보장 범위 명시 |
| P20-new-1 | 발견 | `app.*` 통일 뒤 `app.notification.platform.smtp.max-concurrency`의 env 형태가 SMTP provider profile 키와 겹쳐 변수 하나가 둘 다 설정. | 완료 | Wave 3 P20 `6ef946a0`: 속성을 `dispatch-concurrency`로(env 키 `APP_NOTIFICATION_PLATFORM_SMTP_DISPATCH_CONCURRENCY` 그대로) | - | - |
| P20-new-2 | 발견 | `StrictSettingsBinder.bindExcept`가 system env까지 filter해 하이픈 든 list/map이 env에서 빈 값으로 바인딩. | 완료 | Wave 3 P20 `68cda1d3` | - | - |
| P20-new-3 | 발견 | sample `application.yml`의 `app.cache.redis.*`(22)·`app.outbound.http.*`(15)·`app.messaging.kafka.brokers`·`app.notification.{email,slack}.provider`가 아무것도 바인딩하지 않는 죽은 키. | 남음 | Wave 3 P20 결과 파일 | W5 | 삭제(동작하지 않는 옵션은 키를 두지 않는다) |
| P20-new-4 | 발견 | `src/README.md` "Outbound HTTP client" yaml 예시가 존재하지 않는 트리를 설명. | 남음 | Wave 3 P20 결과 파일 | W5 | 현재 `app.httpclient.*` 트리로 |
| P20-new-5 | 발견 | persistence-jpa 설정이 `app.persistence-jpa`·`app.persistence`·`app.jpa-platform`·`app.lock`·`app.outbox`·`app.realtime.replay`·`app.runtime.migration-on-startup`에 흩어짐 — capability 규칙 4(prefix 하나) 위반. | 남음 | Wave 3 P20 결과 파일 | W4 | 결정 36: 플랫폼 키를 `app.persistence-jpa.*`로 모으고 AGENTS.md 규칙 4에 공유 기능 트리 예외. P33 |
| P20-new-6 | 발견 | graphql strict 바인딩 뒤 빈 `deployment-mode`가 이름 붙은 거절 대신 일반 바인딩 오류로 먼저 실패. | 완료 | Wave 3 P20 `12c61a99` | - | - |
@@ -77,3 +77,14 @@ cd src
- 커밋 메시지 끝: `Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>`
- 보고는 짧게: 변경 / 확인(실행한 명령과 결과 수치) / 남은 문제. Wave마다 Slack `#clean-architecture-전체`(channel_id `C0BCN9H9ABH`)에 작업 브리핑(`:memo: *[작업] …*` 템플릿, 둘째 줄에 `repo`·`branch`·시각).
- 사용자는 작은 증분마다 확인받는 것보다 검증까지 끝낸 큰 덩어리를 선호한다. 진짜로 사용자 몫인 결정만 묻는다(여러 개면 한 번에 모아서, 기본값 추천과 함께). 사용량 한도가 있으니 에이전트는 묶음을 크게 잡아 수를 줄인다(지난번 구현 에이전트는 묶음당 약 36만~72만 토큰).
---
## 일시 정지 상태 (2026-09-29, 사용량 때문에 사용자 요청으로 멈춤)
- 브랜치 HEAD `8bbf1bf0`까지 커밋됨(Wave 3 라운드 1–4의 P30까지). **main 작업 트리에 P20(설정 prefix `app.*`) 통합분이 미커밋으로 있다** — P30과의 충돌 11건과 남은 옛 키 3파일은 이미 풀었다. 재개하면 먼저 Wave 3 종료 검증을 다시 돌리고(로그 scratchpad `w3-*.log`, 멈출 때 실행 중이었음), 통과하면 `docs/capability-status.md`에 `W3-P20.md` 「제거된 코드」 행을 붙이고 커밋 `review-followup Wave 3`.
- 현황표: 완료 606 · 해당 없음 133 · 보류 5 · 남음 173(W4 63 · W5 110). 소유자 결정은 36번까지(현황표 「소유자 결정」 절).
- 계획서 「Wave 4」에 묶음 P31–P34가 있다. P31·P32·P34는 도중에 멈췄다 — worktree 브랜치에 커밋이 남아 있다(`worktree-agent-a3ab3a38ced388a02` P31 4커밋+미커밋 1, `worktree-agent-a37868fbf0dfdeb96` P32 3커밋, `worktree-agent-aa5edcaae241be106` P34 2커밋+미커밋 4, 기준 `8bbf1bf0`). 결과 파일은 scratchpad `W4-P3x.md`(있으면). 재개 시: 각 브랜치 커밋을 확인하고 새 에이전트에 "이 브랜치에서 이어서"로 넘기거나 커밋분만 cherry-pick 후 남은 행으로 새 묶음. P33은 아직 시작 전(P20-new-5 결정 36 포함).
- Wave 3 통합 때 반복된 문제: worktree에서 통과하고 main에서만 실패하는 테스트(저장소 파일을 읽으면서 test inputs 미선언) — P34가 전수 점검 중이었다.
- Wave 3 종료 검증 결과(멈춘 뒤 끝남): mainCheck·architecture/public path/env/config·integration·integrationCheck·build-logic·optional-platforms·releaseCheck/qualificationBootJar 통과, **check·ci 실패** — 재개하면 먼저 원인 확인(로그 scratchpad `w3-check.log`):
> Task :adapter:outbound:persistence-jpa:spotlessJavaCheck FAILED
+1 -1
View File
@@ -3,7 +3,7 @@
Scope: the `:grpc-advanced:*` family. Everything here is off by default and stays off until a
deployment names it. Nothing in this family ships in a runtime composition today.
Flags are `ca-skeleton.grpc.advanced.<capability>.enabled`. `GrpcAdvancedCapability` owns the list of
Flags are `app.grpc.advanced.<capability>.enabled`. `GrpcAdvancedCapability` owns the list of
capability names; `GrpcAdvancedSupportMatrix` owns their current grades.
---
+4 -4
View File
@@ -5,8 +5,8 @@ empty — so nothing here fires in production yet. It is written now because the
the ones an on-call cannot work out from first principles at three in the morning, and shipping the
behaviour before the runbook means the first person to meet one is doing that.
Configuration lives under `ca-skeleton.grpc.platform.*` and is bound by `GrpcPlatformProperties`.
The platform does not start unless `ca-skeleton.grpc.platform.enabled=true`.
Configuration lives under `app.grpc.platform.*` and is bound by `GrpcPlatformProperties`.
The platform does not start unless `app.grpc.platform.enabled=true`.
---
@@ -94,7 +94,7 @@ capacity on nothing.
means the bottleneck is admission, not the work.
2. Check which saturation counter is moving — executor, channel or flow control. They look identical
in a latency graph and have different fixes.
3. Raising `ca-skeleton.grpc.platform.executor-queue-capacity` defers the problem; it does not remove
3. Raising `app.grpc.platform.executor-queue-capacity` defers the problem; it does not remove
it. `GrpcExecutorProfile` refuses a queue above ten thousand for that reason.
---
@@ -138,5 +138,5 @@ construction rather than redacted.
- **Reflection in production.** `GrpcReflectionMode.defaultFor` returns `DISABLED` for `STAGE` and
`PROD`. Reflection publishes the whole schema to anyone who can open a connection.
- **Every advanced capability.** See `docs/runbooks/grpc-advanced-capabilities.md`.
- **The platform itself.** `ca-skeleton.grpc.platform.enabled` defaults to false, and every `:grpc:*`
- **The platform itself.** `app.grpc.platform.enabled` defaults to false, and every `:grpc:*`
leaf is build-only in the registry.
+2 -2
View File
@@ -42,7 +42,7 @@ status: stub
### Step 2 — 임시 격리
- 일시 실패는 자동 backoff 재시도(30s × 2^(attempt-1) + jitter, max attempts 3)가 동작 — 즉시 수동 개입 불필요
- broker 장기 다운이면 DEAD 전이 누적 전에 broker 회복을 우선 (max attempts 소진 시 `runbook://outbox/dead-letter`로 이관)
- relay 자체를 멈춰야 하면 `ca-skeleton.outbox.relay-enabled=false`로 스케줄러 비활성 (이벤트는 outbox 테이블에 안전하게 보존됨 — 유실 없음)
- relay 자체를 멈춰야 하면 `app.outbox.relay-enabled=false`로 스케줄러 비활성 (이벤트는 outbox 테이블에 안전하게 보존됨 — 유실 없음)
## 3. Diagnosis
@@ -62,7 +62,7 @@ status: stub
- 단기: broker 회복 후 backoff 만료 시 자동 재발행 — `outcome=PUBLISHED` 회복 확인
- IN_FLIGHT orphan(claim 후 crash)은 in-flight-timeout(기본 PT5M) 경과 후 자동 재claim — at-least-once이므로 중복 발행 가능, consumer dedupe(idempotencyKey)가 흡수
- 장기: `ca-skeleton.outbox.poll-interval`/`batch-size` 조정, broker 가용성 SLA 점검, 빈발 event_type의 payload 계약 검토
- 장기: `app.outbox.poll-interval`/`batch-size` 조정, broker 가용성 SLA 점검, 빈발 event_type의 payload 계약 검토
## 5. Escalation
+1 -1
View File
@@ -1,5 +1,5 @@
# feature-security-operational-baseline D5 — deny-by-default public path snapshot.
# SSOT: ca-skeleton.security.public-paths default in app-bootstrap/src/main/resources/config/security.yml
# SSOT: app.security.public-paths default in app-bootstrap/src/main/resources/config/security.yml
# -> SecurityConfig permitAll(); anyRequest authenticated. An operator's own SECURITY_PUBLIC_PATHS
# overrides it at run time and is outside this snapshot.
# Update only after review with: ./gradlew verifyPublicPathSnapshot -PapprovePublicPathChange
@@ -236,7 +236,7 @@ W1·W2 잔여 21행(Wave 2.5)은 P14로 1라운드에서 함께 처리한다. W1
- [x] 라운드 1: P14 · P15 · P16 · P17 → 통합·검증·커밋 (2026-09-28, `9fdb0af4`·`c4a066df`·라운드 1 마무리 커밋)
- [x] 라운드 2: P18 · P24 · P25 · P26 → 통합·검증·커밋 (2026-09-28~29, `4fb7943b`·`64a83918`·`25de288a`·`423a29c4`)
- [x] 라운드 3: P19 · P21 · P22 · P23 · P27 · P28 → 통합·검증·커밋 (P23은 P18과 겹치지 않아 라운드 2 자리에서 먼저 시작)
- [ ] 라운드 4: P20 · P30 (P29는 라운드 3에서 끝남) → 통합 → Wave 종료 검증 → 현황표·「제거된 코드」 갱신 → 커밋 `review-followup Wave 3`
- [x] 라운드 4: P20 · P30 (P29는 라운드 3에서 끝남) → 통합 → Wave 종료 검증 → 현황표·「제거된 코드」 갱신 → 커밋 `review-followup Wave 3`
## Wave 4 — 테스트 규칙 (상세 Task: 2026-09-29)
@@ -248,7 +248,7 @@ Wave 3 통합에서 세 번 같은 유형이 드러났다(ORCH-new-2·4·5): 저
|---|---|---|
| P31 | 포트 계약 kit 1 — core·messaging·Redis·멱등성·lock·lease(결정 8): 포트 소유 leaf testFixtures에 추상 `XxxContract`, fake와 실제 어댑터가 함께 | `TEST-F-1` `TEST-F-5` `TEST-FAKE-패턴` `TEST-이행-3` `TEST-FAKE-RedisCommandGateway` `TEST-FAKE-MessagePublisher` `TEST-FAKE-IdempotencyStorePort` `X1-#3a` `X1-#3d` `X1-#6d` `X1-지도-테스트 fake / contract kit` `S6-lead-2` `S7-lead-4` `S2-lead-2` `X1-#6b` `S1-#3b` `P9-new-3` `G-2b` |
| P32 | 포트 계약 kit 2 — object storage·fileserver·notification·sample(Mockito 포트 대역 제거) | `TEST-FAKE-ObjectControlStore` `TEST-FAKE-UploadApplicationService` `TEST-FAKE-DeliveryProjectionStorePort` `X1-#3b` `X1-#3c` `TEST-이행-2b` `P14-new-5` `D-4` `C-10 / S2-#11` `P4-new-1` |
| P33 | 약한 단언·테스트 형식·빠진 테스트, AGENTS.md 규칙 2(DisplayName 한국어), 계층별 모범 모듈 | `TEST-F-2a` `TEST-F-2d` `TEST-F-2e` `TEST-F-2f` `TEST-F-2g` `TEST-F-3` `TEST-F-6` `TEST-F-7` `TEST-F-8` `TEST-이행-4` `TEST-이행-5` `TEST-DisplayName` `J-4c` `J-9b` `H-3d` `I-2c` `I-3c` `E-3c` `E-4c` `E-6a` `F-1d` `F-5d` `K-5b` `L-4a` `L-7b` `S6-#15` `O-3d` |
| P33 | 약한 단언·테스트 형식·빠진 테스트, AGENTS.md 규칙 2(DisplayName 한국어), 계층별 모범 모듈 | `TEST-F-2a` `TEST-F-2d` `TEST-F-2e` `TEST-F-2f` `TEST-F-2g` `TEST-F-3` `TEST-F-6` `TEST-F-7` `TEST-F-8` `TEST-이행-4` `TEST-이행-5` `TEST-DisplayName` `J-4c` `J-9b` `H-3d` `I-2c` `I-3c` `E-3c` `E-4c` `E-6a` `F-1d` `F-5d` `K-5b` `L-4a` `L-7b` `S6-#15` `O-3d` `P20-new-5`(결정 36) |
| P34 | 불안정 테스트·test inputs 누락 전수·compose lane 실제 실행 | `ORCH-new-1` `P25-new-1` `ORCH-new-2` `ORCH-new-4` `NEW-1b` `P11-new-3` `P13-new-2` |
- [ ] P31 · P32 · P33 · P34 병렬(4개) → 통합·Wave 종료 검증 → 커밋 `review-followup Wave 4`
@@ -272,3 +272,4 @@ W5 행 전부, `docs/study/*.ko.md`·`docs/*-superpowers-package/`를 현재 코
현황표: 859행(812 + 후속 발견 47) 중 완료 370 · 해당 없음 112 · 보류 2 · 남음 375(W1 4 · W2 17 · W3 196 · W4 57 · W5 101). 묶음별 결과 파일은 `docs/reviews/2026-09-25-review-followup-results/P*.md`, 에이전트 지시·스크립트는 `docs/superpowers/plans/2026-09-25-review-followup-tools/`.
작업 환경 이동(2026-09-28): 노트북 부하로 이후 작업은 desktop(`ssh desktop`, `/home/donghyeon/workspace/chat-gpt-container/codebase/clean-architecture-backend-template`)에서 이어 한다. 재개 프롬프트: `docs/reviews/2026-09-28-review-followup-resume-prompt.md`.
| 3 | 완료 (2026-09-29) | `check`(leaf 전체) · `ci verifyDependencyLocks --warning-mode=fail`(714 task) · `mainCheck` · architecture/public path/env keys/config contract · bootstrap·sample integration/system · `integrationCheck`(로컬 통과 — NEW-16 해결) · build-logic test · optional-platforms `ci` · `releaseCheck` · `qualificationBootJar` 통과. 라운드마다 통합 게이트 + 벤더 lane(PG·Kafka·Redis topology·MinIO·Mongo) | 라운드 커밋 `9fdb0af4`…`8bbf1bf0` + `review-followup Wave 3` | 소유자 결정 28–36 추가. 통합 때 드러난 문제: worktree에서 통과·main에서 실패하는 테스트(test inputs 미선언, ORCH-new-2·4·5), 합친 뒤 의미 충돌(P18 × P25, P20 × P30) |
+3 -3
View File
@@ -12,9 +12,9 @@ nothing happens until service code calls it.
| Capability | How it is turned on | What it adds | Tests |
| --- | --- | --- | --- |
| MVC virtual threads | `backend.web.advanced.mvc-virtual-threads.enabled=true` (+ admission limit and budgets) | A different scheduling model for **every** request | `MvcAdvancedConfigurationTest`, `VirtualThreadAdmissionTest` |
| NDJSON streaming | `backend.web.advanced.ndjson.enabled=true` | Long-lived connections | `MvcStreamingWiringTest`, `WebStreamCoreTest`, `WebStreamDrainTest`, `WebStreamingSoakIT` |
| JSON text sequences (RFC 7464) | `backend.web.advanced.json-sequence.enabled=true` | Long-lived connections | as NDJSON |
| MVC virtual threads | `app.web.advanced.mvc-virtual-threads.enabled=true` (+ admission limit and budgets) | A different scheduling model for **every** request | `MvcAdvancedConfigurationTest`, `VirtualThreadAdmissionTest` |
| NDJSON streaming | `app.web.advanced.ndjson.enabled=true` | Long-lived connections | `MvcStreamingWiringTest`, `WebStreamCoreTest`, `WebStreamDrainTest`, `WebStreamingSoakIT` |
| JSON text sequences (RFC 7464) | `app.web.advanced.json-sequence.enabled=true` | Long-lived connections | as NDJSON |
| JSON Merge Patch (RFC 7396) | library — `JsonMergePatchApplier` | Partial documents | `JsonMergePatchApplierTest` |
| JSON Patch (RFC 6902) | library — `JsonPatchApplier` | Arbitrary pointers | `JsonPatchApplierTest` |
| CBOR / XML codecs | library — `WebCborMapperFactory`, `WebXmlMapperFactory`; the backend jar is the service's choice | A second decoder | `WebCodecMapperTest`, `WebCodecSecurityTest` |
+2 -2
View File
@@ -80,7 +80,7 @@ equivalents in `application-core/idempotency` — `IdempotencyStorePort`, `Idemp
`IdempotencyScopeDigest` and `RequestFingerprint` — with two provider stores behind the port:
PostgreSQL (`PostgreSqlIdempotencyStore` in `persistence-jpa`) and Redis
(`RedisIdempotencyStoreAdapter` in `cache-redis`), selected by
`ca-skeleton.capabilities.idempotency.provider`. That is what the design's Tasks 39 and 40 ask for,
`app.capabilities.idempotency.provider`. That is what the design's Tasks 39 and 40 ask for,
built in the direction the registry permits.
Implementing the design's SPI literally would have required
@@ -141,7 +141,7 @@ answers in the `Envelope` contract, and a caller's operation that is not theirs
(`OPERATION_NOT_FOUND`) as one that does not exist. The controller and the three use cases are
assembled by `app-bootstrap` `autoconfigure/operation/DurableOperationConfig`, which turns on only
when the operation store and the outbox append port exist (the JPA module with
`ca-skeleton.outbox.enabled=true`). Submit and cancel need `operation:submit` / `operation:cancel`
`app.outbox.enabled=true`). Submit and cancel need `operation:submit` / `operation:cancel`
(`security.yml` grants both to the `operation-submit` role).
The state machine is enforced three times on purpose: in `DurableOperation`'s constructor, in the
+1 -1
View File
@@ -65,7 +65,7 @@ is one series per resource.
## Cross-origin requests fail only in the browser
The API answers correctly and the browser refuses the response. CORS is `CorsSettings`
(`ca-skeleton.cors.*`) applied by `SecurityConfig`. Startup refuses two settings:
(`app.cors.*`) applied by `SecurityConfig`. Startup refuses two settings:
- `APP_SECURITY_CORS_ENABLED=true` with no origin.
- `*` in `APP_SECURITY_CORS_ORIGINS` with `APP_SECURITY_CORS_ALLOW_CREDENTIALS=true` — no browser
+1 -1
View File
@@ -5,7 +5,7 @@ framing exists, but since the WebFlux twin was removed (2026-09-25) nothing on t
writes it.
The writers ship in `adapter:inbound:web-advanced` (opt-in, switched by
`backend.web.advanced.ndjson.enabled` / `backend.web.advanced.json-sequence.enabled`); no default
`app.web.advanced.ndjson.enabled` / `app.web.advanced.json-sequence.enabled`); no default
composition carries them.
## Three outcomes, not two
+1 -1
View File
@@ -20,7 +20,7 @@ The profile ships in `adapter:inbound:web-advanced`, which no default compositio
that leaf to the service, then two settings — the profile refuses to be constructed with only one:
```yaml
backend:
app:
web:
advanced:
mvc-virtual-threads:
+2 -2
View File
@@ -1,6 +1,6 @@
user default off
user ca-skeleton-application on >fixture-application sanitize-payload ~prod:* resetchannels &prod:* -@all +@connection +@pubsub +@transaction +@read +@write +@string +@hash +@list +@set +@sortedset +@bitmap +@hyperloglog +@geo +@stream -keys -flushdb -flushall -shutdown -debug -sort -sort_ro -smembers -randomkey -migrate -swapdb -select +cluster|slots +cluster|shards +cluster|nodes +cluster|info +cluster|myid
user ca-skeleton-application-advanced on >fixture-advanced sanitize-payload ~prod:* resetchannels &prod:* -@all +@read +@write +@string +@hash +@list +@set +@sortedset +@bitmap +@hyperloglog +@geo +@stream +@pubsub +@transaction +time +evalsha +evalsha_ro +script|load +script|exists +fcall +fcall_ro -keys -flushdb -flushall -shutdown -debug -eval -eval_ro -smembers -sort -sort_ro -randomkey -migrate -swapdb -select +cluster|slots +cluster|shards +cluster|nodes +cluster|info +cluster|myid
user ca-skeleton-application on >fixture-application sanitize-payload ~prod:* resetchannels -@all +@connection +@transaction +@read +@write +@string +@hash +@list +@set +@sortedset +@bitmap +@hyperloglog +@geo +@stream -keys -flushdb -flushall -shutdown -debug -sort -sort_ro -smembers -randomkey -migrate -swapdb -select +cluster|slots +cluster|shards +cluster|nodes +cluster|info +cluster|myid
user ca-skeleton-application-advanced on >fixture-advanced sanitize-payload ~prod:* resetchannels -@all +@read +@write +@string +@hash +@list +@set +@sortedset +@bitmap +@hyperloglog +@geo +@stream +@transaction +time +evalsha +evalsha_ro +script|load +script|exists +fcall +fcall_ro -keys -flushdb -flushall -shutdown -debug -eval -eval_ro -smembers -sort -sort_ro -randomkey -migrate -swapdb -select +cluster|slots +cluster|shards +cluster|nodes +cluster|info +cluster|myid
user ca-skeleton-raw-gateway on >fixture-raw sanitize-payload ~prod:* resetchannels -@all +smembers +sort +sort_ro
user ca-skeleton-admin-readonly on >fixture-admin ~* resetchannels -@all +info +dbsize +time +lastsave +memory|usage +memory|stats +slowlog|get +slowlog|len +latency|latest +latency|history +client|list +client|info +command|info +command|docs +command|count +command|getkeysandflags +config|get +acl|dryrun +acl|whoami +cluster|info +cluster|slots +cluster|shards +cluster|nodes +object|encoding +object|freq +object|idletime +pubsub|channels +pubsub|numsub +pubsub|shardchannels +xinfo|stream +xinfo|groups +xinfo|consumers +function|list +function|stats +cluster|keyslot +cluster|myid
user ca-skeleton-replication on >fixture-replication ~* resetchannels -@all +psync +replconf +ping
+1 -1
View File
@@ -1 +1 @@
user ca-skeleton-application-advanced on >fixture-advanced sanitize-payload ~prod:* resetchannels &prod:* -@all +@read +@write +@string +@hash +@list +@set +@sortedset +@bitmap +@hyperloglog +@geo +@stream +@pubsub +@transaction +time +evalsha +evalsha_ro +script|load +script|exists +fcall +fcall_ro -keys -flushdb -flushall -shutdown -debug -eval -eval_ro -smembers -sort -sort_ro -randomkey -migrate -swapdb -select
user ca-skeleton-application-advanced on >fixture-advanced sanitize-payload ~prod:* resetchannels -@all +@read +@write +@string +@hash +@list +@set +@sortedset +@bitmap +@hyperloglog +@geo +@stream +@transaction +time +evalsha +evalsha_ro +script|load +script|exists +fcall +fcall_ro -keys -flushdb -flushall -shutdown -debug -eval -eval_ro -smembers -sort -sort_ro -randomkey -migrate -swapdb -select
+1 -1
View File
@@ -1 +1 @@
user ca-skeleton-application on >fixture-application sanitize-payload ~prod:* resetchannels &prod:* -@all +@connection +@pubsub +@transaction +@read +@write +@string +@hash +@list +@set +@sortedset +@bitmap +@hyperloglog +@geo +@stream -keys -flushdb -flushall -shutdown -debug -sort -sort_ro -smembers -randomkey -migrate -swapdb -select
user ca-skeleton-application on >fixture-application sanitize-payload ~prod:* resetchannels -@all +@connection +@transaction +@read +@write +@string +@hash +@list +@set +@sortedset +@bitmap +@hyperloglog +@geo +@stream -keys -flushdb -flushall -shutdown -debug -sort -sort_ro -smembers -randomkey -migrate -swapdb -select
+2 -2
View File
@@ -72,8 +72,8 @@ def configuration_properties() -> list[str]:
"""Every leaf property under `app.httpclient`, nested and dynamic blocks included.
Read from the environment-field manifest rather than from the record source. The manifest is
derived from `HttpClientPlatformSettings` by `HttpClientEnvironmentKeys` and held to it in both
directions by `HttpClientPlatformEnvManifestTest`, so it cannot drift from the code; parsing the
derived from `HttpClientPlatformSettings` and held to it in both directions by
`HttpClientPlatformEnvManifestTest`, so it cannot drift from the code; parsing the
record here a second time, with a regex, could only agree with it by luck. The previous version
of this function did exactly that and saw eighteen top-level names, which is why a nested pool,
timeout or TLS setting could be added and documented nowhere.
+3 -3
View File
@@ -270,7 +270,7 @@ Kafka 없이 기동합니다. 나머지 기능은 `app-bootstrap/build.gradle`
| 외부 HTTP 호출 | `runtimeOnly project(':adapter:outbound:httpclient')` | client 하나 이상: `APP_HTTPCLIENT_CLIENTS_0_*` (`docs/httpclient/env-fields.yaml`) | `APP_HTTPCLIENT_ENABLED` |
| 객체 저장소 (S3/로컬) | `runtimeOnly project(':adapter:outbound:objectstorage')` | 전용 env 없음. `app.object-storage.*`의 provider·destination을 `application-*.yml`에 적음 | `app.object-storage.enabled` |
| WebSocket (STOMP push) | `runtimeOnly project(':adapter:inbound:websocket')` | 없음(기본값으로 동작). origin은 `APP_WEBSOCKET_ALLOWED_ORIGINS` | `APP_WEBSOCKET_ENABLED` |
| gRPC 서버 | `runtimeOnly project(':adapter:inbound:grpc')` | `CA_SKELETON_GRPC_SERVER_ALLOW_INSECURE_LOCAL=true` (지금 받는 listener가 loopback plaintext뿐) | `ca-skeleton.grpc.server.enabled` |
| gRPC 서버 | `runtimeOnly project(':adapter:inbound:grpc')` | `APP_GRPC_SERVER_ALLOW_INSECURE_LOCAL=true` (지금 받는 listener가 loopback plaintext뿐) | `app.grpc.server.enabled` |
| 파일 서버 (JPA를 함께 가져옴) | `runtimeOnly project(':composition:fileserver-platform')` | `APP_FILESERVER_PLATFORM_STORAGE_ROOT`, `APP_FILESERVER_PLATFORM_OBSERVABILITY_FINGERPRINT_KEY`, 그리고 `FileAccessPolicyPort` bean 또는 `APP_FILESERVER_PLATFORM_SECURITY_ACCESS_POLICY=role-based` | `APP_FILESERVER_PLATFORM_ENABLED` |
| 알림 (SMTP/SES, JPA 줄도 필요) | `runtimeOnly project(':composition:notification-platform')` | 키 8종과 각 `_ID`(`APP_NOTIFICATION_PLATFORM_*_KEY`, `*_KEY_ID`), 쓰는 provider 설정(`APP_NOTIFICATION_PLATFORM_SMTP_*` 또는 `_SES_*`) | `APP_NOTIFICATION_PLATFORM_ENABLED` |
@@ -338,7 +338,7 @@ Kafka·Redis는 위 "기능 추가 = 한 줄 + env" 표의 한 줄로 고릅니
현재 canonical activation은 다음 두 설정 트리만 사용합니다.
```yaml
ca-skeleton:
app:
capabilities:
http-client:
expected-state: DISABLED
@@ -539,7 +539,7 @@ docker compose -f docker-compose.yml -f docker-compose.local.yml up -d --wait db
컨테이너로 띄우는 `scripts/bootstrap-local.sh` 경로는 이 호스트 포트를 쓰지 않습니다. compose 가 app
컨테이너의 `APP_DATASOURCE_URL` 을 내부 네트워크 주소 `jdbc:postgresql://db:5432/...` 로 덮어씁니다.
`ca-skeleton.persistence.vendor`(`postgresql` | `h2`)가 어느 vendor 구성을 조립할지 고르는 단일
`app.persistence.vendor`(`postgresql` | `h2`)가 어느 vendor 구성을 조립할지 고르는 단일
스위치입니다. 값이 둘 중 하나가 아니면 기동이 실패하고, prod 에서 `h2` 이거나 datasource URL 이
`jdbc:h2:` 이면 `PersistenceVendorProdSafetyValidator` 가 기동을 거부합니다(env 로 덮어써도 동일).
+1 -1
View File
@@ -57,7 +57,7 @@ required-class 검사가 이 테스트가 조용히 사라지는 것을 막는
- feature-agnostic: `classpath:graphql/**` 스키마와 모든 `@Controller` `@QueryMapping`/
`@MutationMapping` 을 generic 하게 합성한다. **WorkLog 등 구체 기능을 이름으로 알지 않는다.**
- 있으면 켜진다: 서비스가 `app-bootstrap/build.gradle`에 `runtimeOnly project(':adapter:inbound:graphql')`
한 줄로 고른다(기본 조립에는 없다). `APP_GRAPHQL_ENABLED=false`(`backend.graphql.enabled`)만 끄며,
한 줄로 고른다(기본 조립에는 없다). `APP_GRAPHQL_ENABLED=false`(`app.graphql.enabled`)만 끄며,
그때 `GraphQlActivationEnvironmentPostProcessor`가 Spring GraphQL 자동 설정도 exclude 한다.
실제 활성화 시 composition root 는 인증/인가·CORS 정책과 GraphiQL/introspection 운영 설정을 함께
명시해야 한다.
+3 -3
View File
@@ -102,18 +102,18 @@ gRPC 와 달리 spring-graphql / graphql-java 는 Spring Boot BOM 이 관리한
버전 명시도, 모듈 스코프 platform import 도 필요 없다 — `build.gradle` 은 BOM-managed 좌표만
선언하고, per-module `gradle.lockfile` 이 strict locking 으로 정확한 버전을 고정한다.
## 설정 — 프레임워크 `spring.graphql.*` + 플랫폼 `backend.graphql.*`
## 설정 — 프레임워크 `spring.graphql.*` + 플랫폼 `app.graphql.*`
전송 계층 설정(path, graphiql, introspection, schema location)은 프레임워크 `spring.graphql.*`
가 소유한다. composition-root `application.yml` 에서 설정하며 모듈별 `yml` 은 없다.
플랫폼 정책은 `spring.graphql.*` 로 표현할 수 없다 — cost/page 한계, preparsed
캐시 경계, 관측 label 로 허용할 operation 이름은 전부 이 leaf 의 결정이다. 그래서
`GraphQlPlatformProperties` 가 **`backend.graphql`** prefix 로 `@ConfigurationProperties` 를
`GraphQlPlatformProperties` 가 **`app.graphql`** prefix 로 `@ConfigurationProperties` 를
바인딩한다(`spring.graphql.platform.*` 이 아니다 — 그 prefix 는 존재한 적이 없다).
```yaml
backend:
app:
graphql:
deployment-mode: PRODUCTION_INTERNAL
validation-policy-version: v1 # preparsed 캐시 키의 일부
+2 -2
View File
@@ -67,8 +67,8 @@ dependencies {
// GraphQlPlatformProperties is a @ConfigurationProperties binding, so this leaf owes the
// repository-wide processor parity gate (`verifyConfigurationPropertiesProcessor`) a metadata
// declaration — an adopter configuring backend.graphql.* gets IDE completion and validation
// from the generated metadata rather than from prose. (The prefix is `backend.graphql`; this
// declaration — an adopter configuring app.graphql.* gets IDE completion and validation
// from the generated metadata rather than from prose. (The prefix is `app.graphql`; this
// comment used to say `spring.graphql.platform.*`, which never existed.)
// A raw request body can only be capped before something decodes it, and on a servlet stack the
@@ -16,11 +16,13 @@ import org.springframework.boot.context.properties.bind.Binder;
import org.springframework.boot.env.YamlPropertySourceLoader;
import org.springframework.core.Ordered;
import org.springframework.core.env.ConfigurableEnvironment;
import org.springframework.core.env.EnumerablePropertySource;
import org.springframework.core.env.MapPropertySource;
import org.springframework.core.env.PropertySource;
import org.springframework.core.io.ClassPathResource;
/**
* Everything about {@code backend.graphql} that has to be settled before a bean exists.
* Everything about {@code app.graphql} that has to be settled before a bean exists.
*
* <p>An environment post-processor rather than a bean: these are the failures that have to arrive
* first, naming what an operator set, instead of surfacing as a bean somewhere downstream that
@@ -28,8 +30,8 @@ import org.springframework.core.io.ClassPathResource;
*
* <ol>
* <li><b>Contribute the module's settings.</b> {@code config/graphql.yml} maps the {@code
* APP_GRAPHQL_*} variables onto {@code backend.graphql.*}, which an environment variable
* cannot reach directly. Added last, so a deployment's own value always wins, and loaded only
* APP_GRAPHQL_*} variables onto {@code app.graphql.*}, which an environment variable cannot
* reach directly. Added last, so a deployment's own value always wins, and loaded only
* because the module is on the classpath — removing the module removes its settings.
* <li><b>Refuse a retired safety key.</b> See {@link GraphQlRetiredSafetyAxis}.
* <li><b>Give the framework's console flags the platform's defaults.</b> Spring Boot answers
@@ -37,7 +39,7 @@ import org.springframework.core.io.ClassPathResource;
* the lowest precedence, so an operator who sets either key still wins and still gets
* validated.
* <li><b>Make an explicit off complete.</b> The module is on when present. {@code
* backend.graphql.enabled=false} turns off {@link GraphQlRootAutoConfiguration}, but Spring
* app.graphql.enabled=false} turns off {@link GraphQlRootAutoConfiguration}, but Spring
* GraphQL's own auto-configurations would still publish {@code /graphql} from the schema on
* the classpath. So an explicit off appends them — read from Boot's own registration file,
* not listed here — to {@code spring.autoconfigure.exclude}, after whatever the deployment
@@ -52,7 +54,9 @@ public class GraphQlActivationEnvironmentPostProcessor
static final String RESOURCE = "config/graphql.yml";
private static final String SWITCH = "backend.graphql.enabled";
private static final String SWITCH = "app.graphql.enabled";
private static final String DEPLOYMENT_MODE = "app.graphql.deployment-mode";
private static final String EXCLUDE = "spring.autoconfigure.exclude";
@@ -63,8 +67,8 @@ public class GraphQlActivationEnvironmentPostProcessor
private static final Map<String, String> FRAMEWORK_DEFAULTS =
Map.of(
"spring.graphql.schema.introspection.enabled",
"backend.graphql.console.introspection-enabled",
"spring.graphql.graphiql.enabled", "backend.graphql.console.graphiql-enabled");
"app.graphql.console.introspection-enabled",
"spring.graphql.graphiql.enabled", "app.graphql.console.graphiql-enabled");
private static final String SOURCE_NAME = "graphQlPlatformConsoleDefaults";
@@ -76,6 +80,7 @@ public class GraphQlActivationEnvironmentPostProcessor
new YamlPropertySourceLoader()
.load("graphql-defaults (" + RESOURCE + ")", new ClassPathResource(RESOURCE, classLoader))
.forEach(environment.getPropertySources()::addLast);
withoutAnUnsetDeploymentMode(environment);
} catch (IOException failure) {
throw new UncheckedIOException("failed to read " + RESOURCE, failure);
}
@@ -100,6 +105,37 @@ public class GraphQlActivationEnvironmentPostProcessor
}
}
/**
* Drops the shipped {@code deployment-mode} entry when it resolves to nothing.
*
* <p>The mode has no default on purpose, and {@link GraphQlDeploymentMode} refuses a missing one
* by name. The shipped file maps it as {@code ${APP_GRAPHQL_DEPLOYMENT_MODE:}}, though, and under
* strict binding ({@code ignoreUnknownFields = false}) a present-but-empty key is reported as
* "left unbound" before the named refusal can run. An unset mode is therefore made absent, not
* empty; a mode set anywhere — environment, {@code .env}, system property, file — still binds.
*/
private static void withoutAnUnsetDeploymentMode(ConfigurableEnvironment environment) {
String mode = environment.getProperty(DEPLOYMENT_MODE);
if (mode != null && !mode.isBlank()) {
return;
}
for (PropertySource<?> source : environment.getPropertySources().stream().toList()) {
if (source.getName().startsWith("graphql-defaults")
&& source instanceof EnumerablePropertySource<?> enumerable
&& enumerable.containsProperty(DEPLOYMENT_MODE)) {
Map<String, Object> kept = new LinkedHashMap<>();
for (String name : enumerable.getPropertyNames()) {
if (!DEPLOYMENT_MODE.equals(name)) {
kept.put(name, enumerable.getProperty(name));
}
}
environment
.getPropertySources()
.replace(source.getName(), new MapPropertySource(source.getName(), kept));
}
}
}
private static void excludeTheFrameworkAutoConfigurations(
ConfigurableEnvironment environment, ClassLoader classLoader) {
List<String> excluded =
@@ -43,7 +43,7 @@ public enum GraphQlDeploymentMode {
/** What an operator has to set, and where, when the mode is missing. */
public static final String REQUIRED_MESSAGE =
"backend.graphql.deployment-mode is required while backend.graphql.enabled=true; set "
"app.graphql.deployment-mode is required while app.graphql.enabled=true; set "
+ "APP_GRAPHQL_DEPLOYMENT_MODE to one of LOCAL, DEV, PRODUCTION_INTERNAL, "
+ "PRODUCTION_PUBLIC";
@@ -36,8 +36,8 @@ class GraphQlExecutionChainConfiguration {
* application knowledge: a deny-by-default skeleton policy would answer nothing and adopters
* would replace it with an allow-all one, while an allow-by-default policy shipped into
* production would be an unauthorized endpoint. So a development mode gets the permissive default
* that lets the schema be explored, and a production {@code backend.graphql.deployment-mode}
* refuses to start without an explicit policy.
* that lets the schema be explored, and a production {@code app.graphql.deployment-mode} refuses
* to start without an explicit policy.
*/
@Bean
@ConditionalOnMissingBean
@@ -49,7 +49,7 @@ class GraphQlExecutionChainConfiguration {
if (properties.production()) {
throw new GraphQlPlatformConfigurationException(
List.of(
"backend.graphql.deployment-mode="
"app.graphql.deployment-mode="
+ properties.deploymentMode()
+ " requires an explicit GraphQlAuthorizationPolicy bean; the platform has no "
+ "application coordinates to authorize on its own"));
@@ -23,14 +23,20 @@ import org.springframework.boot.context.properties.bind.DefaultValue;
* @param deploymentMode the single safety posture; required, see {@link GraphQlDeploymentMode}
* @param console query console and introspection exposure
* @param limits page and cost ceilings
* @param validationPolicyVersion part of the preparsed cache key
* @param observedOperationNames operation names allowed as an observation label
* @param enabled the capability switch. Conditions read the property; it is a component so that the
* tree binds with unknown keys refused ({@code ignoreUnknownFields = false}) — a misspelt
* {@code app.graphql.*} key fails startup instead of leaving a default limit in force
*/
@ConfigurationProperties("backend.graphql")
@ConfigurationProperties(prefix = "app.graphql", ignoreUnknownFields = false)
public record GraphQlPlatformSettings(
GraphQlDeploymentMode deploymentMode,
@DefaultValue Console console,
@DefaultValue Limits limits,
@DefaultValue("v1") String validationPolicyVersion,
@DefaultValue Set<String> observedOperationNames) {
@DefaultValue Set<String> observedOperationNames,
@DefaultValue("true") boolean enabled) {
public GraphQlPlatformSettings {
// Belt and braces for programmatic construction: the binder honours @DefaultValue, but this
@@ -109,34 +115,45 @@ public record GraphQlPlatformSettings(
Console.disabled(),
Limits.defaults(),
"v1",
Set.of());
Set.of(),
true);
}
/** Returns a copy running under a different deployment mode. */
public GraphQlPlatformSettings withDeploymentMode(GraphQlDeploymentMode mode) {
return new GraphQlPlatformSettings(
mode, console, limits, validationPolicyVersion, observedOperationNames);
mode, console, limits, validationPolicyVersion, observedOperationNames, enabled);
}
/** Returns a copy with GraphiQL enabled or disabled. */
public GraphQlPlatformSettings withGraphiqlEnabled(boolean enabled) {
return withConsole(new Console(enabled, console.introspectionEnabled()));
public GraphQlPlatformSettings withGraphiqlEnabled(boolean graphiql) {
return withConsole(new Console(graphiql, console.introspectionEnabled()));
}
/** Returns a copy with introspection enabled or disabled. */
public GraphQlPlatformSettings withIntrospectionEnabled(boolean enabled) {
return withConsole(new Console(console.graphiqlEnabled(), enabled));
public GraphQlPlatformSettings withIntrospectionEnabled(boolean introspection) {
return withConsole(new Console(console.graphiqlEnabled(), introspection));
}
/** Returns a copy with a different console policy. */
public GraphQlPlatformSettings withConsole(Console replacement) {
return new GraphQlPlatformSettings(
deploymentMode, replacement, limits, validationPolicyVersion, observedOperationNames);
deploymentMode,
replacement,
limits,
validationPolicyVersion,
observedOperationNames,
enabled);
}
/** Returns a copy with different limits. */
public GraphQlPlatformSettings withLimits(Limits replacement) {
return new GraphQlPlatformSettings(
deploymentMode, console, replacement, validationPolicyVersion, observedOperationNames);
deploymentMode,
console,
replacement,
validationPolicyVersion,
observedOperationNames,
enabled);
}
}
@@ -74,7 +74,7 @@ public final class GraphQlPlatformStartupValidator {
}
throw new GraphQlPlatformConfigurationException(
List.of(
"backend.graphql.deployment-mode="
"app.graphql.deployment-mode="
+ mode
+ " is not permitted in the "
+ runtime.profileName()
@@ -121,7 +121,7 @@ public final class GraphQlPlatformStartupValidator {
problems.add(
"client policy maxPageSize ("
+ runtime.clientPolicy().maxPageSize()
+ ") exceeds backend.graphql.limits.maximum-page-size ("
+ ") exceeds app.graphql.limits.maximum-page-size ("
+ runtime.properties().limits().maximumPageSize()
+ ")");
}
@@ -130,14 +130,14 @@ public final class GraphQlPlatformStartupValidator {
problems.add(
"client policy maxComplexity ("
+ runtime.clientPolicy().maxComplexity()
+ ") exceeds backend.graphql.limits.maximum-complexity ("
+ ") exceeds app.graphql.limits.maximum-complexity ("
+ runtime.properties().limits().maximumComplexity()
+ ")");
}
if (runtime.clientPolicy().introspectionAllowed()
&& !runtime.properties().console().introspectionEnabled()) {
problems.add(
"client policy allows introspection while backend.graphql.console.introspection-enabled "
"client policy allows introspection while app.graphql.console.introspection-enabled "
+ "is false; one of the two is not what the operator configured");
}
@@ -148,7 +148,7 @@ public final class GraphQlPlatformStartupValidator {
&& runtime.frameworkIntrospectionEnabled()
!= runtime.properties().console().introspectionEnabled()) {
problems.add(
"backend.graphql.console.introspection-enabled ("
"app.graphql.console.introspection-enabled ("
+ runtime.properties().console().introspectionEnabled()
+ ") contradicts spring.graphql.schema.introspection.enabled ("
+ runtime.frameworkIntrospectionEnabled()
@@ -157,7 +157,7 @@ public final class GraphQlPlatformStartupValidator {
if (runtime.frameworkGraphiqlEnabled() != null
&& runtime.frameworkGraphiqlEnabled() != runtime.properties().console().graphiqlEnabled()) {
problems.add(
"backend.graphql.console.graphiql-enabled ("
"app.graphql.console.graphiql-enabled ("
+ runtime.properties().console().graphiqlEnabled()
+ ") contradicts spring.graphql.graphiql.enabled ("
+ runtime.frameworkGraphiqlEnabled()
@@ -41,7 +41,7 @@ class GraphQlRequestContextConfiguration {
/**
* The limits applied to a request when the adopter has not registered a client policy.
*
* <p>Page size, complexity and introspection follow {@code backend.graphql.*}; the rest are the
* <p>Page size, complexity and introspection follow {@code app.graphql.*}; the rest are the
* calibration starting points from the design. An adopter replaces this bean rather than editing
* a table of constants.
*/
@@ -76,7 +76,7 @@ class GraphQlRequestContextConfiguration {
if (properties.production()) {
throw new GraphQlPlatformConfigurationException(
List.of(
"backend.graphql.deployment-mode="
"app.graphql.deployment-mode="
+ properties.deploymentMode()
+ " requires an explicit GraphQlPrincipalResolver bean; production refuses "
+ "anonymous callers, so the anonymous default would reject every request"));
@@ -8,11 +8,11 @@ import org.springframework.core.env.Environment;
/**
* The two retired safety keys, refused rather than ignored (GQL-INT-002).
*
* <p>{@code backend.graphql.production} and {@code backend.graphql.environment} are no longer
* record components, and Spring's binder ignores unknown keys by default. Left alone, an operator
* who sets the key they have always set gets a clean startup and a silently different safety
* posture — which is a worse outcome than the split-brain being fixed, because the old
* configuration at least did something.
* <p>{@code app.graphql.production} and {@code app.graphql.environment} are no longer record
* components, and Spring's binder ignores unknown keys by default. Left alone, an operator who sets
* the key they have always set gets a clean startup and a silently different safety posture — which
* is a worse outcome than the split-brain being fixed, because the old configuration at least did
* something.
*
* <p>Checked whether or not the replacement is also set. A deployment carrying both is exactly the
* state that produced the defect: two keys that can disagree about one question, with the winner
@@ -26,10 +26,10 @@ public final class GraphQlRetiredSafetyAxis {
private static final Map<String, String> RETIRED_KEYS =
Map.of(
"backend.graphql.production",
"app.graphql.production",
"it drove anonymous principal handling and allow-by-default authorization while the "
+ "environment enum drove introspection, so the two could disagree",
"backend.graphql.environment",
"app.graphql.environment",
"it drove introspection and GraphiQL while the production boolean drove authorization, "
+ "so the two could disagree");
@@ -42,7 +42,7 @@ public final class GraphQlRetiredSafetyAxis {
* @return the problems; empty when the capability switch is off or no retired key is present
*/
public static List<String> problems(Environment environment) {
if (!"true".equalsIgnoreCase(environment.getProperty("backend.graphql.enabled", "true"))) {
if (!"true".equalsIgnoreCase(environment.getProperty("app.graphql.enabled", "true"))) {
return List.of();
}
List<String> problems = new ArrayList<>();
@@ -53,7 +53,7 @@ public final class GraphQlRetiredSafetyAxis {
+ " is retired ("
+ RETIRED_KEYS.get(key)
+ "). Remove it and set APP_GRAPHQL_DEPLOYMENT_MODE "
+ "(backend.graphql.deployment-mode) to one of LOCAL, DEV, PRODUCTION_INTERNAL, "
+ "(app.graphql.deployment-mode) to one of LOCAL, DEV, PRODUCTION_INTERNAL, "
+ "PRODUCTION_PUBLIC.");
}
}
@@ -9,17 +9,17 @@ import org.springframework.context.annotation.Import;
* The one place that decides whether this application exposes a GraphQL endpoint.
*
* <p>On when present: an application that selected this module wants GraphQL, so a missing {@code
* backend.graphql.enabled} means on. Absent from the build is the ordinary off state.
* app.graphql.enabled} means on. Absent from the build is the ordinary off state.
*
* <p>An inbound adapter's off contract is not only about beans. Spring GraphQL publishes {@code
* /graphql} from its own auto-configuration, with no project bean involved, so an explicit {@code
* backend.graphql.enabled=false} also excludes those auto-configurations — see {@link
* app.graphql.enabled=false} also excludes those auto-configurations — see {@link
* GraphQlActivationEnvironmentPostProcessor}. A bean-inventory assertion cannot see a route the
* framework published.
*/
@AutoConfiguration
@ConditionalOnProperty(
prefix = "backend.graphql",
prefix = "app.graphql",
name = "enabled",
havingValue = "true",
matchIfMissing = true)
@@ -2,8 +2,8 @@
#
# 이 모듈이 직접 읽는다(GraphQlActivationEnvironmentPostProcessor). 우선순위는 가장 낮아서
# application*.yml, 환경 변수, 시스템 속성이 항상 이긴다. 환경 변수는 APP_GRAPHQL_* 이고 설정 트리는
# backend.graphql.* 라서, 이 파일이 둘을 잇는다. 이 모듈을 빌드에서 빼면 이 파일도 함께 빠진다.
backend:
# app.graphql.* 라서, 이 파일이 둘을 잇는다. 이 모듈을 빌드에서 빼면 이 파일도 함께 빠진다.
app:
graphql:
# true | false — 모듈이 있으면 켜진다. false 로 끄면 Spring GraphQL 자신의 자동 설정도 빠져서
# /graphql 이 열리지 않는다.
@@ -29,4 +29,4 @@ backend:
info:
capabilities:
graphql:
enabled: ${backend.graphql.enabled:true}
enabled: ${app.graphql.enabled:true}
@@ -51,10 +51,10 @@ import org.springframework.web.cors.UrlBasedCorsConfigurationSource;
// The transport is off unless a deployment asks for it, so a boundary qualification has to
// ask too. Before the master switch existed this test passed by accident: the endpoint was
// published from the classpath.
"backend.graphql.enabled=true",
"app.graphql.enabled=true",
// Required alongside the switch: the safety posture has no default, because the defaults it
// used to have were a production enum and a non-production boolean that disagreed.
"backend.graphql.deployment-mode=LOCAL",
"app.graphql.deployment-mode=LOCAL",
"spring.graphql.graphiql.enabled=false",
"spring.graphql.schema.introspection.enabled=false",
"spring.graphql.schema.locations=classpath:graphql-qualification-no-discovery/",
@@ -62,7 +62,7 @@ import org.springframework.stereotype.Controller;
webEnvironment = SpringBootTest.WebEnvironment.RANDOM_PORT,
properties = {
// The transport is off unless a deployment asks for it, so an execution-path test has to ask.
"backend.graphql.enabled=true",
"app.graphql.enabled=true",
"spring.graphql.graphiql.enabled=false",
// The schema permits introspection at the deployment level, and the two flags agree so the
// contradiction check stays quiet. What refuses the query is the anonymous profile's client
@@ -73,12 +73,12 @@ import org.springframework.stereotype.Controller;
"spring.graphql.schema.additional-files="
+ "classpath:graphql/skeleton.graphqls,"
+ "classpath:graphql-platform/platform.graphqls",
"backend.graphql.deployment-mode=LOCAL",
"backend.graphql.console.introspection-enabled=true",
"backend.graphql.limits.maximum-page-size=100",
"app.graphql.deployment-mode=LOCAL",
"app.graphql.console.introspection-enabled=true",
"app.graphql.limits.maximum-page-size=100",
// Small enough that one nested selection exceeds it, large enough that the allowed queries do
// not. The unregistered-coordinate default weight is what both sides are measured against.
"backend.graphql.limits.maximum-complexity=100"
"app.graphql.limits.maximum-complexity=100"
})
@AutoConfigureTestRestTemplate
class GraphQlPlatformExecutionPathTest {
@@ -27,8 +27,7 @@ class GraphQlBatchLoaderWiringTest {
private final ApplicationContextRunner runner =
new ApplicationContextRunner()
.withConfiguration(AutoConfigurations.of(GraphQlRootAutoConfiguration.class))
.withPropertyValues(
"backend.graphql.enabled=true", "backend.graphql.deployment-mode=LOCAL");
.withPropertyValues("app.graphql.enabled=true", "app.graphql.deployment-mode=LOCAL");
@Test
@DisplayName("the platform supplies the whole batch chain when GraphQL is on")
@@ -1,8 +1,10 @@
package dev.caskeleton.adapter.inbound.graphql.autoconfigure;
import static org.assertj.core.api.Assertions.assertThat;
import static org.assertj.core.api.Assertions.catchThrowable;
import java.util.Arrays;
import org.junit.jupiter.api.DisplayName;
import org.junit.jupiter.api.Test;
import org.springframework.boot.WebApplicationType;
import org.springframework.boot.autoconfigure.EnableAutoConfiguration;
@@ -49,6 +51,18 @@ class GraphQlCapabilityActivationTest {
}
}
@Test
@DisplayName("mode를 정하지 않으면 strict 바인딩 오류가 아니라 이름 붙은 거절로 기동이 실패한다")
void anUnsetModeIsRefusedByNameRatherThanAsAnUnboundKey() {
// when
Throwable thrown = catchThrowable(() -> start().close());
// then
assertThat(thrown)
.hasStackTraceContaining("app.graphql.deployment-mode is required")
.hasStackTraceContaining("APP_GRAPHQL_DEPLOYMENT_MODE");
}
private static ConfigurableApplicationContext start(String... arguments) {
return new SpringApplicationBuilder(Application.class)
.web(WebApplicationType.NONE)
@@ -26,9 +26,9 @@ import org.springframework.core.env.StandardEnvironment;
/**
* One deployment mode decides every production-safety behaviour (GQL-INT-002).
*
* <p>There used to be two axes. {@code backend.graphql.production} defaulted to {@code false} and
* {@code backend.graphql.environment} defaulted to {@code PRODUCTION_PUBLIC}, in the same record,
* so the shipped default was a deployment that called itself internet-facing production while every
* <p>There used to be two axes. {@code app.graphql.production} defaulted to {@code false} and
* {@code app.graphql.environment} defaulted to {@code PRODUCTION_PUBLIC}, in the same record, so
* the shipped default was a deployment that called itself internet-facing production while every
* protection that read the boolean behaved as if it were not. That is not a display inconsistency:
* anonymous principal handling, allow-by-default authorization, and part of request protection read
* only the boolean, so the enum could say production while the endpoint answered as development.
@@ -42,7 +42,7 @@ class GraphQlDeploymentModeContractTest {
private final ApplicationContextRunner runner =
new ApplicationContextRunner()
.withConfiguration(AutoConfigurations.of(GraphQlRootAutoConfiguration.class))
.withPropertyValues("backend.graphql.enabled=true");
.withPropertyValues("app.graphql.enabled=true");
@ParameterizedTest
@EnumSource(GraphQlDeploymentMode.class)
@@ -81,15 +81,32 @@ class GraphQlDeploymentModeContractTest {
assertThat(failure)
.isInstanceOf(GraphQlPlatformConfigurationException.class)
.hasMessageContaining("backend.graphql.deployment-mode=LOCAL")
.hasMessageContaining("app.graphql.deployment-mode=LOCAL")
.hasMessageContaining("[PRODUCTION_INTERNAL, PRODUCTION_PUBLIC]");
}
@Test
@DisplayName("app.graphql 아래 오타 키는 기동을 실패시킨다")
void aMisspelledGraphQlKeyFailsStartup() {
// when
runner
.withPropertyValues(
"app.graphql.deployment-mode=LOCAL",
"spring.profiles.active=local",
"app.graphql.limits.maximum-page-sise=5")
.run(
context -> {
// then
assertThat(context).hasFailed();
assertThat(context.getStartupFailure()).hasStackTraceContaining("maximum-page-sise");
});
}
@Test
@DisplayName("the context refuses a mode that does not match its runtime profile")
void theContextRefusesAModeForAnotherRuntime() {
runner
.withPropertyValues("backend.graphql.deployment-mode=DEV", "spring.profiles.active=local")
.withPropertyValues("app.graphql.deployment-mode=DEV", "spring.profiles.active=local")
.run(
context ->
assertThat(context)
@@ -107,7 +124,7 @@ class GraphQlDeploymentModeContractTest {
assertThat(new GraphQlPlatformStartupValidator().problems(settings))
.as("a message that does not name the property sends an operator reading")
.anySatisfy(problem -> assertThat(problem).contains("backend.graphql.deployment-mode"));
.anySatisfy(problem -> assertThat(problem).contains("app.graphql.deployment-mode"));
}
@Test
@@ -144,7 +161,7 @@ class GraphQlDeploymentModeContractTest {
}
@ParameterizedTest
@ValueSource(strings = {"backend.graphql.production", "backend.graphql.environment"})
@ValueSource(strings = {"app.graphql.production", "app.graphql.environment"})
@DisplayName("a retired safety key is a migration error naming the replacement")
void aRetiredKeyIsAMigrationError(String retired) {
List<String> problems =
@@ -163,8 +180,8 @@ class GraphQlDeploymentModeContractTest {
GraphQlRetiredSafetyAxis.problems(
environmentWith(
Map.of(
"backend.graphql.deployment-mode", "LOCAL",
"backend.graphql.environment", "PRODUCTION_PUBLIC")));
"app.graphql.deployment-mode", "LOCAL",
"app.graphql.environment", "PRODUCTION_PUBLIC")));
assertThat(problems)
.as(
@@ -175,8 +192,7 @@ class GraphQlDeploymentModeContractTest {
@Test
@DisplayName("with the master off the retired keys are not read at all")
void withTheMasterOffTheNamespaceIsNotValidated() {
ConfigurableEnvironment environment =
environmentWith(Map.of("backend.graphql.production", "true"));
ConfigurableEnvironment environment = environmentWith(Map.of("app.graphql.production", "true"));
environment.getPropertySources().addFirst(off());
assertThat(GraphQlRetiredSafetyAxis.problems(environment))
@@ -193,14 +209,14 @@ class GraphQlDeploymentModeContractTest {
.hasFailed()
.getFailure()
.rootCause()
.hasMessageContaining("backend.graphql.deployment-mode"));
.hasMessageContaining("app.graphql.deployment-mode"));
}
@Test
@DisplayName("a development mode gets the permissive authorization default")
void aDevelopmentModeGetsThePermissiveDefault() {
runner
.withPropertyValues("backend.graphql.deployment-mode=LOCAL")
.withPropertyValues("app.graphql.deployment-mode=LOCAL")
.run(
context -> {
assertThat(context).hasNotFailed();
@@ -216,7 +232,7 @@ class GraphQlDeploymentModeContractTest {
@DisplayName("a production mode refuses the permissive default, naming the mode")
void aProductionModeRefusesThePermissiveDefault(String mode) {
runner
.withPropertyValues("backend.graphql.deployment-mode=" + mode)
.withPropertyValues("app.graphql.deployment-mode=" + mode)
.run(
context ->
assertThat(context)
@@ -231,7 +247,7 @@ class GraphQlDeploymentModeContractTest {
@DisplayName("production 은 익명 principal resolver 로 떠서 모든 요청을 거부하는 대신 기동을 거부한다")
void aProductionModeRefusesTheAnonymousPrincipalResolver(String mode) {
runner
.withPropertyValues("backend.graphql.deployment-mode=" + mode)
.withPropertyValues("app.graphql.deployment-mode=" + mode)
.withBean(
GraphQlAuthorizationPolicy.class, () -> GraphQlAuthorizationPolicy.builder().build())
.run(
@@ -247,7 +263,7 @@ class GraphQlDeploymentModeContractTest {
@DisplayName("production 은 인가 정책과 principal resolver 를 둘 다 주면 기동한다")
void aProductionModeWithAnExplicitPolicyAndResolverStarts() {
runner
.withPropertyValues("backend.graphql.deployment-mode=PRODUCTION_INTERNAL")
.withPropertyValues("app.graphql.deployment-mode=PRODUCTION_INTERNAL")
.withBean(
GraphQlAuthorizationPolicy.class, () -> GraphQlAuthorizationPolicy.builder().build())
.withBean(GraphQlPrincipalResolver.class, () -> request -> Optional.empty())
@@ -258,7 +274,7 @@ class GraphQlDeploymentModeContractTest {
@DisplayName("the mode reaches the interceptor that decides what an anonymous caller may do")
void theModeReachesTheRequestPath() {
runner
.withPropertyValues("backend.graphql.deployment-mode=LOCAL")
.withPropertyValues("app.graphql.deployment-mode=LOCAL")
.run(
context -> {
GraphQlPlatformSettings settings = context.getBean(GraphQlPlatformSettings.class);
@@ -304,7 +320,7 @@ class GraphQlDeploymentModeContractTest {
@DisplayName("the platform's console setting is what the framework is given")
void thePlatformConsoleSettingReachesTheFramework() {
ConfigurableEnvironment environment =
environmentWith(Map.of("backend.graphql.console.introspection-enabled", "true"));
environmentWith(Map.of("app.graphql.console.introspection-enabled", "true"));
new GraphQlActivationEnvironmentPostProcessor().postProcessEnvironment(environment, null);
@@ -331,12 +347,12 @@ class GraphQlDeploymentModeContractTest {
environment.getPropertySources().addFirst(source("contract", properties));
environment
.getPropertySources()
.addLast(source("master", Map.of("backend.graphql.enabled", "true")));
.addLast(source("master", Map.of("app.graphql.enabled", "true")));
return environment;
}
private static MapPropertySource off() {
return source("off", Map.of("backend.graphql.enabled", "false"));
return source("off", Map.of("app.graphql.enabled", "false"));
}
private static MapPropertySource source(String name, Map<String, String> properties) {
@@ -27,8 +27,7 @@ class GraphQlObservationWiringTest {
AutoConfigurations.of(
GraphQlRootAutoConfiguration.class, GraphQlObservationAutoConfiguration.class))
.withBean(ObservationRegistry.class, ObservationRegistry::create)
.withPropertyValues(
"backend.graphql.enabled=true", "backend.graphql.deployment-mode=LOCAL");
.withPropertyValues("app.graphql.enabled=true", "app.graphql.deployment-mode=LOCAL");
@Test
void theFrameworkInstrumentationResolvesThePlatformConvention() {
@@ -41,8 +41,7 @@ class GraphQlPlatformAutoConfigurationTest {
// it is testing the missing-mode refusal rather than whatever it meant to test. That
// refusal has its own cases in GraphQlDeploymentModeContractTest. Later property values
// override earlier ones, so a case that needs a different mode simply names it.
.withPropertyValues(
"backend.graphql.enabled=true", "backend.graphql.deployment-mode=LOCAL");
.withPropertyValues("app.graphql.enabled=true", "app.graphql.deployment-mode=LOCAL");
@Test
void aContextWithOnlyTheDeploymentModeSetBootsWithSafeDefaults() {
@@ -81,7 +80,7 @@ class GraphQlPlatformAutoConfigurationTest {
@Test
void productionRefusesToStartWithoutAnExplicitAuthorizationPolicy() {
runner
.withPropertyValues("backend.graphql.deployment-mode=PRODUCTION_INTERNAL")
.withPropertyValues("app.graphql.deployment-mode=PRODUCTION_INTERNAL")
.run(
context ->
assertThat(context)
@@ -116,7 +115,7 @@ class GraphQlPlatformAutoConfigurationTest {
@Test
void aClientPolicyLooserThanTheConfiguredCeilingIsRejectedAtStartup() {
runner
.withPropertyValues("backend.graphql.limits.maximum-page-size=25")
.withPropertyValues("app.graphql.limits.maximum-page-size=25")
.withBean(GraphQlClientPolicy.class, () -> GraphQlClientPolicy.defaults(500, 10_000, false))
.run(
context ->
@@ -124,7 +123,7 @@ class GraphQlPlatformAutoConfigurationTest {
.hasFailed()
.getFailure()
.rootCause()
.hasMessageContaining("exceeds backend.graphql.limits.maximum-page-size"));
.hasMessageContaining("exceeds app.graphql.limits.maximum-page-size"));
}
@Test
@@ -137,8 +136,7 @@ class GraphQlPlatformAutoConfigurationTest {
runner
.withPropertyValues(
"backend.graphql.deployment-mode=LOCAL",
"backend.graphql.console.introspection-enabled=true")
"app.graphql.deployment-mode=LOCAL", "app.graphql.console.introspection-enabled=true")
.withBean(GraphQlProperties.class, () -> framework)
.run(
context ->
@@ -156,7 +154,7 @@ class GraphQlPlatformAutoConfigurationTest {
framework.getGraphiql().setEnabled(true);
runner
.withPropertyValues("backend.graphql.deployment-mode=LOCAL")
.withPropertyValues("app.graphql.deployment-mode=LOCAL")
.withBean(GraphQlProperties.class, () -> framework)
.run(
context ->
@@ -174,8 +172,7 @@ class GraphQlPlatformAutoConfigurationTest {
runner
.withPropertyValues(
"backend.graphql.deployment-mode=LOCAL",
"backend.graphql.console.introspection-enabled=true")
"app.graphql.deployment-mode=LOCAL", "app.graphql.console.introspection-enabled=true")
.withBean(GraphQlProperties.class, () -> framework)
.run(context -> assertThat(context).hasNotFailed());
}
@@ -200,19 +197,19 @@ class GraphQlPlatformAutoConfigurationTest {
String metadata = readClasspathResource("META-INF/spring-configuration-metadata.json");
assertThat(metadata)
.contains("backend.graphql.deployment-mode")
.doesNotContain("backend.graphql.production")
.doesNotContain("backend.graphql.environment")
.contains("backend.graphql.console.graphiql-enabled")
.contains("backend.graphql.console.introspection-enabled")
.contains("backend.graphql.limits.maximum-page-size")
.contains("backend.graphql.limits.maximum-complexity")
.contains("app.graphql.deployment-mode")
.doesNotContain("app.graphql.production")
.doesNotContain("app.graphql.environment")
.contains("app.graphql.console.graphiql-enabled")
.contains("app.graphql.console.introspection-enabled")
.contains("app.graphql.limits.maximum-page-size")
.contains("app.graphql.limits.maximum-complexity")
// Keys that only failed the boot or were echoed back unused were removed (2026-09-25).
.doesNotContain("backend.graphql.cursor.key-ids")
.doesNotContain("backend.graphql.unsupported.multipart-upload")
.doesNotContain("app.graphql.cursor.key-ids")
.doesNotContain("app.graphql.unsupported.multipart-upload")
// Servlet-only template (D4): no reactive profile to choose, no event loop to bridge off.
.doesNotContain("backend.graphql.execution-profile")
.doesNotContain("backend.graphql.unbridged-blocking-resolvers");
.doesNotContain("app.graphql.execution-profile")
.doesNotContain("app.graphql.unbridged-blocking-resolvers");
}
private static GraphQlProperties frameworkDefaults() {
@@ -35,8 +35,7 @@ class GraphQlPolicyRequestPathTest {
private final ApplicationContextRunner runner =
new ApplicationContextRunner()
.withConfiguration(AutoConfigurations.of(GraphQlRootAutoConfiguration.class))
.withPropertyValues(
"backend.graphql.enabled=true", "backend.graphql.deployment-mode=LOCAL");
.withPropertyValues("app.graphql.enabled=true", "app.graphql.deployment-mode=LOCAL");
@Test
@DisplayName("the policies that are wired reach the request path through the instrumentation")
@@ -39,8 +39,7 @@ class GraphQlPlatformWebInterceptorRequestBoundsTest {
private final ApplicationContextRunner runner =
new ApplicationContextRunner()
.withConfiguration(AutoConfigurations.of(GraphQlRootAutoConfiguration.class))
.withPropertyValues(
"backend.graphql.enabled=true", "backend.graphql.deployment-mode=LOCAL")
.withPropertyValues("app.graphql.enabled=true", "app.graphql.deployment-mode=LOCAL")
.withBean(
GraphQlClientPolicy.class,
() ->
@@ -44,8 +44,7 @@ class GraphQlRequestStageErrorMappingTest {
private final ApplicationContextRunner runner =
new ApplicationContextRunner()
.withConfiguration(AutoConfigurations.of(GraphQlRootAutoConfiguration.class))
.withPropertyValues(
"backend.graphql.enabled=true", "backend.graphql.deployment-mode=LOCAL");
.withPropertyValues("app.graphql.enabled=true", "app.graphql.deployment-mode=LOCAL");
@Test
@DisplayName("principal resolver 가 던진 ApiErrorCarrier 는 요청 단계에서도 자기 code 로 나간다")
+3 -3
View File
@@ -38,11 +38,11 @@ Package root: `dev.caskeleton.adapter.inbound.grpc`.
- 프로덕션 feature RPC 를 스켈레톤에 두는 것 — health/reflection 표면만 (web 의
`HealthcheckController` 와 동일 원칙).
## Config knobs (`ca-skeleton.grpc.server.*`)
## Config knobs (`app.grpc.server.*`)
타입드 `@ConfigurationProperties` 만 두고, 값은 composition-root `application.yml` 에 산다
(모듈별 `yml` 없음). prefix가 `ca-skeleton.grpc`가 아니라 `.server`까지인 이유: 바인딩이
`ignoreUnknownFields = false`라서, 부모 prefix를 쓰면 벤더 starter의 `ca-skeleton.grpc.platform.*`
(모듈별 `yml` 없음). prefix가 `app.grpc`가 아니라 `.server`까지인 이유: 바인딩이
`ignoreUnknownFields = false`라서, 부모 prefix를 쓰면 벤더 starter의 `app.grpc.platform.*`
키가 들어오는 순간 기동이 실패한다(`GrpcSafeActivationTest`).
| key | default | 의미 |
+3 -3
View File
@@ -22,10 +22,10 @@
```
이 줄을 지우면 서버, 포트, 설정 바인딩이 모두 함께 빠진다.
- 켜기와 끄기: 모듈이 있으면 켜진다. `ca-skeleton.grpc.server.enabled=false`로만 끈다.
- 켜기와 끄기: 모듈이 있으면 켜진다. `app.grpc.server.enabled=false`로만 끈다.
- 필요한 설정: 전용 `APP_*` env 키는 registry에 없다. 켜진 상태에서 지금 구현이 받는 listener는
loopback plaintext뿐이라 `ca-skeleton.grpc.server.allow-insecure-local=true`(env
`CA_SKELETON_GRPC_SERVER_ALLOW_INSECURE_LOCAL=true`)를 명시해야 기동한다. 선택: `ca-skeleton.grpc.server.port`
loopback plaintext뿐이라 `app.grpc.server.allow-insecure-local=true`(env
`APP_GRPC_SERVER_ALLOW_INSECURE_LOCAL=true`)를 명시해야 기동한다. 선택: `app.grpc.server.port`
(기본 9090), `.bind-address`(기본 127.0.0.1), `.reflection-enabled`, `.shutdown-grace-seconds`.
## 왜 self-managed Netty 인가
@@ -11,8 +11,8 @@ import org.springframework.context.annotation.Bean;
/**
* Wires the gRPC transport machinery. Registered through this module's own {@code
* AutoConfiguration.imports}: on when the module is present, off with {@code
* ca-skeleton.grpc.server.enabled=false}. The plaintext listener still refuses to start unless
* {@code ca-skeleton.grpc.server.allow-insecure-local=true} is explicit.
* app.grpc.server.enabled=false}. The plaintext listener still refuses to start unless {@code
* app.grpc.server.allow-insecure-local=true} is explicit.
*
* <p>All collaborators are plain objects composed here, mirroring the clean DI style used across
* the skeleton. A feature service is exposed only when its {@link BindableService} bean carries
@@ -20,7 +20,7 @@ import org.springframework.context.annotation.Bean;
*/
@AutoConfiguration
@ConditionalOnProperty(
prefix = "ca-skeleton.grpc.server",
prefix = "app.grpc.server",
name = "enabled",
havingValue = "true",
matchIfMissing = true)
@@ -10,11 +10,11 @@ import org.springframework.boot.context.properties.ConfigurationProperties;
import org.springframework.validation.annotation.Validated;
/**
* gRPC server settings bound from {@code ca-skeleton.grpc.server.*}. Typed configuration only (no
* gRPC server settings bound from {@code app.grpc.server.*}. Typed configuration only (no
* per-module {@code yml}); values live in the composition-root {@code application.yml}, matching
* the ca-skeleton config convention. See README for the self-managed-Netty rationale.
*/
@ConfigurationProperties(prefix = "ca-skeleton.grpc.server", ignoreUnknownFields = false)
@ConfigurationProperties(prefix = "app.grpc.server", ignoreUnknownFields = false)
@Validated
public class GrpcServerSettings {
@@ -70,11 +70,11 @@ class GrpcP1BoundaryWireTest {
new ApplicationContextRunner()
.withUserConfiguration(GrpcServerConfig.class, FeatureConfiguration.class)
.withPropertyValues(
"ca-skeleton.grpc.server.enabled=true",
"ca-skeleton.grpc.server.port=0",
"ca-skeleton.grpc.server.bind-address=127.0.0.1",
"ca-skeleton.grpc.server.allow-insecure-local=true",
"ca-skeleton.grpc.server.reflection-enabled=false");
"app.grpc.server.enabled=true",
"app.grpc.server.port=0",
"app.grpc.server.bind-address=127.0.0.1",
"app.grpc.server.allow-insecure-local=true",
"app.grpc.server.reflection-enabled=false");
@Test
void missingAuthenticationMetadataIsRejectedWithAStableContract() {
@@ -42,15 +42,14 @@ class GrpcSafeActivationTest {
@Test
void presentWithoutTheSwitchIsOnAndStillRefusesAnImplicitPlaintextListener() {
contextRunner
.withPropertyValues("ca-skeleton.grpc.server.port=0")
.withPropertyValues("app.grpc.server.port=0")
.run(context -> assertRootCauseContains(context.getStartupFailure(), "insecure"));
}
@Test
void switchedOffCreatesNoGrpcRuntimeBeansOrListener() {
contextRunner
.withPropertyValues(
"ca-skeleton.grpc.server.enabled=false", "ca-skeleton.grpc.server.port=0")
.withPropertyValues("app.grpc.server.enabled=false", "app.grpc.server.port=0")
.run(
context -> {
assertThat(context).doesNotHaveBean(GrpcServerSettings.class);
@@ -63,8 +62,7 @@ class GrpcSafeActivationTest {
@Test
void enabledTransportRequiresAnExplicitLocalInsecureOverride() {
contextRunner
.withPropertyValues(
"ca-skeleton.grpc.server.enabled=true", "ca-skeleton.grpc.server.port=0")
.withPropertyValues("app.grpc.server.enabled=true", "app.grpc.server.port=0")
.run(context -> assertRootCauseContains(context.getStartupFailure(), "insecure"));
}
@@ -72,10 +70,10 @@ class GrpcSafeActivationTest {
void insecureTransportRejectsANonLoopbackBindAddress() {
contextRunner
.withPropertyValues(
"ca-skeleton.grpc.server.enabled=true",
"ca-skeleton.grpc.server.port=0",
"ca-skeleton.grpc.server.bind-address=0.0.0.0",
"ca-skeleton.grpc.server.allow-insecure-local=true")
"app.grpc.server.enabled=true",
"app.grpc.server.port=0",
"app.grpc.server.bind-address=0.0.0.0",
"app.grpc.server.allow-insecure-local=true")
.run(context -> assertRootCauseContains(context.getStartupFailure(), "loopback"));
}
@@ -83,10 +81,10 @@ class GrpcSafeActivationTest {
void enabledTransportRejectsAPortOutsideTheTcpRange() {
contextRunner
.withPropertyValues(
"ca-skeleton.grpc.server.enabled=true",
"ca-skeleton.grpc.server.port=65536",
"ca-skeleton.grpc.server.bind-address=127.0.0.1",
"ca-skeleton.grpc.server.allow-insecure-local=true")
"app.grpc.server.enabled=true",
"app.grpc.server.port=65536",
"app.grpc.server.bind-address=127.0.0.1",
"app.grpc.server.allow-insecure-local=true")
.run(context -> assertRootCauseContains(context.getStartupFailure(), "port"));
}
@@ -94,11 +92,11 @@ class GrpcSafeActivationTest {
void enabledTransportRejectsANegativeShutdownGrace() {
contextRunner
.withPropertyValues(
"ca-skeleton.grpc.server.enabled=true",
"ca-skeleton.grpc.server.port=0",
"ca-skeleton.grpc.server.bind-address=127.0.0.1",
"ca-skeleton.grpc.server.allow-insecure-local=true",
"ca-skeleton.grpc.server.shutdown-grace-seconds=-1")
"app.grpc.server.enabled=true",
"app.grpc.server.port=0",
"app.grpc.server.bind-address=127.0.0.1",
"app.grpc.server.allow-insecure-local=true",
"app.grpc.server.shutdown-grace-seconds=-1")
.run(
context ->
assertRootCauseContains(context.getStartupFailure(), "shutdownGraceSeconds"));
@@ -109,10 +107,10 @@ class GrpcSafeActivationTest {
contextRunner
.withUserConfiguration(FeatureServiceConfiguration.class)
.withPropertyValues(
"ca-skeleton.grpc.server.enabled=true",
"ca-skeleton.grpc.server.port=0",
"ca-skeleton.grpc.server.bind-address=127.0.0.1",
"ca-skeleton.grpc.server.allow-insecure-local=true")
"app.grpc.server.enabled=true",
"app.grpc.server.port=0",
"app.grpc.server.bind-address=127.0.0.1",
"app.grpc.server.allow-insecure-local=true")
.run(context -> assertRootCauseContains(context.getStartupFailure(), "authentication"));
}
@@ -120,9 +118,9 @@ class GrpcSafeActivationTest {
void theVendorPlatformNamespaceBesideTheAdapterDoesNotFailItsStrictBinding() {
contextRunner
.withPropertyValues(
"ca-skeleton.grpc.server.port=0",
"ca-skeleton.grpc.server.allow-insecure-local=true",
"ca-skeleton.grpc.platform.enabled=true")
"app.grpc.server.port=0",
"app.grpc.server.allow-insecure-local=true",
"app.grpc.platform.enabled=true")
.run(context -> assertThat(context).hasNotFailed());
}
@@ -27,10 +27,10 @@ import org.springframework.context.annotation.Configuration;
/**
* Boots the skeleton gRPC machinery in a real Spring context on an ephemeral port ({@code
* ca-skeleton.grpc.server.port=0}) with ZERO feature services and proves it stands up a working
* surface: the {@link GrpcServerRunner} SmartLifecycle starts, the standard {@code grpc.health.v1}
* health service reports SERVING, and v1 server reflection lists the built-in services. A real
* Netty channel exercises the wire, so this is a genuine transport smoke test, not a wiring mock.
* app.grpc.server.port=0}) with ZERO feature services and proves it stands up a working surface:
* the {@link GrpcServerRunner} SmartLifecycle starts, the standard {@code grpc.health.v1} health
* service reports SERVING, and v1 server reflection lists the built-in services. A real Netty
* channel exercises the wire, so this is a genuine transport smoke test, not a wiring mock.
*/
class GrpcServerRunnerBootTest {
@@ -38,11 +38,11 @@ class GrpcServerRunnerBootTest {
new ApplicationContextRunner()
.withUserConfiguration(GrpcServerConfig.class)
.withPropertyValues(
"ca-skeleton.grpc.server.enabled=true",
"ca-skeleton.grpc.server.port=0",
"ca-skeleton.grpc.server.bind-address=127.0.0.1",
"ca-skeleton.grpc.server.allow-insecure-local=true",
"ca-skeleton.grpc.server.reflection-enabled=true");
"app.grpc.server.enabled=true",
"app.grpc.server.port=0",
"app.grpc.server.bind-address=127.0.0.1",
"app.grpc.server.allow-insecure-local=true",
"app.grpc.server.reflection-enabled=true");
@Test
void skeletonServerStartsAndServesHealthAndReflectionWithNoFeatures() {
+3 -3
View File
@@ -24,9 +24,9 @@ registry(`src/config/architecture/modules.json`)의 그 leaf `allowed_dependenci
| 기능 | 켜는 설정 | 만드는 bean |
| --- | --- | --- |
| MVC virtual thread 실행기 + admission 제한 | `backend.web.advanced.mvc-virtual-threads.enabled=true`, `admission-limit`, `database-pool-size`, `outbound-bulkhead`, `admission-wait` | `applicationTaskExecutor`, `VirtualThreadAdmissionGuard`, `VirtualThreadProfile` |
| NDJSON 스트리밍 | `backend.web.advanced.ndjson.enabled=true` | `mvcNdjsonWriter` + 공유 `WebStreamPolicy`·`WebStreamRegistry`·`StreamRecordEncoder` |
| JSON text sequence (RFC 7464) | `backend.web.advanced.json-sequence.enabled=true` | `mvcJsonSequenceWriter` + 같은 공유 bean |
| MVC virtual thread 실행기 + admission 제한 | `app.web.advanced.mvc-virtual-threads.enabled=true`, `admission-limit`, `database-pool-size`, `outbound-bulkhead`, `admission-wait` | `applicationTaskExecutor`, `VirtualThreadAdmissionGuard`, `VirtualThreadProfile` |
| NDJSON 스트리밍 | `app.web.advanced.ndjson.enabled=true` | `mvcNdjsonWriter` + 공유 `WebStreamPolicy`·`WebStreamRegistry`·`StreamRecordEncoder` |
| JSON text sequence (RFC 7464) | `app.web.advanced.json-sequence.enabled=true` | `mvcJsonSequenceWriter` + 같은 공유 bean |
virtual thread 실행기는 `TaskDecorator` bean 을 요구한다(요청 상관관계 전파). 조립 순서는
`TaskExecutionAutoConfiguration` 앞이다. 스트리밍 두 기능은 Boot 의 `JsonMapper` bean 으로 레코드를
@@ -15,7 +15,7 @@ import org.springframework.context.annotation.Import;
@AutoConfiguration
@ConditionalOnWebApplication(type = ConditionalOnWebApplication.Type.SERVLET)
@ConditionalOnProperty(
prefix = "backend.web.advanced.json-sequence",
prefix = "app.web.advanced.json-sequence",
name = "enabled",
havingValue = "true")
@Import(MvcStreamingSupportConfiguration.class)
@@ -14,10 +14,7 @@ import org.springframework.context.annotation.Import;
/** Wires servlet-side NDJSON streaming only. */
@AutoConfiguration
@ConditionalOnWebApplication(type = ConditionalOnWebApplication.Type.SERVLET)
@ConditionalOnProperty(
prefix = "backend.web.advanced.ndjson",
name = "enabled",
havingValue = "true")
@ConditionalOnProperty(prefix = "app.web.advanced.ndjson", name = "enabled", havingValue = "true")
@Import(MvcStreamingSupportConfiguration.class)
public class MvcStreamingExecutorConfiguration {
@@ -33,7 +33,7 @@ import org.springframework.core.task.support.TaskExecutorAdapter;
@EnableConfigurationProperties(VirtualThreadSettings.class)
@ConditionalOnWebApplication(type = ConditionalOnWebApplication.Type.SERVLET)
@ConditionalOnProperty(
prefix = "backend.web.advanced.mvc-virtual-threads",
prefix = "app.web.advanced.mvc-virtual-threads",
name = "enabled",
havingValue = "true")
public class VirtualThreadMvcConfiguration {
@@ -7,14 +7,14 @@ import org.springframework.boot.context.properties.ConfigurationProperties;
import org.springframework.validation.annotation.Validated;
/**
* Settings bound from {@code backend.web.advanced.mvc-virtual-threads.*}.
* Settings bound from {@code app.web.advanced.mvc-virtual-threads.*}.
*
* <p>The downstream budgets are settings rather than something read from the pool at runtime,
* deliberately. Reading them would make the check pass automatically — whatever the pool is, the
* admission limit would be compared against it and the operator would never be asked to think about
* the relationship. Writing them down here is what makes a mismatch visible in a diff.
*/
@ConfigurationProperties(prefix = "backend.web.advanced.mvc-virtual-threads")
@ConfigurationProperties(prefix = "app.web.advanced.mvc-virtual-threads")
@Validated
public class VirtualThreadSettings {
@@ -46,10 +46,10 @@ class MvcAdvancedConfigurationTest {
// thread has no relationship to the submitter.
new WebApplicationContextRunner()
.withPropertyValues(
"backend.web.advanced.mvc-virtual-threads.enabled=true",
"backend.web.advanced.mvc-virtual-threads.admission-limit=8",
"backend.web.advanced.mvc-virtual-threads.database-pool-size=4",
"backend.web.advanced.mvc-virtual-threads.outbound-bulkhead=4")
"app.web.advanced.mvc-virtual-threads.enabled=true",
"app.web.advanced.mvc-virtual-threads.admission-limit=8",
"app.web.advanced.mvc-virtual-threads.database-pool-size=4",
"app.web.advanced.mvc-virtual-threads.outbound-bulkhead=4")
.withUserConfiguration(VirtualThreadMvcConfiguration.class)
.run(context -> assertThat(context).hasFailed());
}
@@ -77,7 +77,7 @@ class MvcAdvancedConfigurationTest {
// Four beans with a dependency chain between them. A cycle or a missing dependency here is not
// visible from any unit test of the classes involved.
runner
.withPropertyValues("backend.web.advanced.ndjson.enabled=true")
.withPropertyValues("app.web.advanced.ndjson.enabled=true")
.withUserConfiguration(MvcStreamingExecutorConfiguration.class)
.run(
context ->
@@ -94,7 +94,7 @@ class MvcAdvancedConfigurationTest {
@DisplayName("JSON sequence has its own activation property")
void jsonSequenceStartsIndependentlyOfNdjson() {
runner
.withPropertyValues("backend.web.advanced.json-sequence.enabled=true")
.withPropertyValues("app.web.advanced.json-sequence.enabled=true")
.withUserConfiguration(MvcJsonSequenceConfiguration.class)
.run(
context ->
@@ -114,10 +114,10 @@ class MvcAdvancedConfigurationTest {
// concurrency bound and replaced it with nothing.
runner
.withPropertyValues(
"backend.web.advanced.mvc-virtual-threads.enabled=true",
"backend.web.advanced.mvc-virtual-threads.admission-limit=100",
"backend.web.advanced.mvc-virtual-threads.database-pool-size=20",
"backend.web.advanced.mvc-virtual-threads.outbound-bulkhead=20")
"app.web.advanced.mvc-virtual-threads.enabled=true",
"app.web.advanced.mvc-virtual-threads.admission-limit=100",
"app.web.advanced.mvc-virtual-threads.database-pool-size=20",
"app.web.advanced.mvc-virtual-threads.outbound-bulkhead=20")
.withUserConfiguration(VirtualThreadMvcConfiguration.class)
.run(
context -> {
@@ -139,10 +139,10 @@ class MvcAdvancedConfigurationTest {
// place — the capability switched on and nothing changed.
runner
.withPropertyValues(
"backend.web.advanced.mvc-virtual-threads.enabled=true",
"backend.web.advanced.mvc-virtual-threads.admission-limit=8",
"backend.web.advanced.mvc-virtual-threads.database-pool-size=4",
"backend.web.advanced.mvc-virtual-threads.outbound-bulkhead=4")
"app.web.advanced.mvc-virtual-threads.enabled=true",
"app.web.advanced.mvc-virtual-threads.admission-limit=8",
"app.web.advanced.mvc-virtual-threads.database-pool-size=4",
"app.web.advanced.mvc-virtual-threads.outbound-bulkhead=4")
.withUserConfiguration(VirtualThreadMvcConfiguration.class)
.run(
context ->
@@ -158,7 +158,7 @@ class MvcAdvancedConfigurationTest {
// will not refresh, which is the intended outcome: a node that accepts every arrival and
// queues it on budgets that did not grow is worse than a node that does not start.
runner
.withPropertyValues("backend.web.advanced.mvc-virtual-threads.enabled=true")
.withPropertyValues("app.web.advanced.mvc-virtual-threads.enabled=true")
.withUserConfiguration(VirtualThreadMvcConfiguration.class)
.run(
context ->
@@ -64,8 +64,8 @@ class WebAdvancedRollbackIT {
// matchIfMissing = true would pass the test above and fail this one.
runner
.withPropertyValues(
"backend.web.advanced.ndjson.enabled=false",
"backend.web.advanced.mvc-virtual-threads.enabled=false")
"app.web.advanced.ndjson.enabled=false",
"app.web.advanced.mvc-virtual-threads.enabled=false")
.withUserConfiguration(
MvcStreamingExecutorConfiguration.class,
MvcJsonSequenceConfiguration.class,
@@ -21,7 +21,7 @@ import org.springframework.context.annotation.Import;
@AutoConfiguration
@ConditionalOnWebApplication(type = ConditionalOnWebApplication.Type.SERVLET)
@ConditionalOnProperty(
prefix = "ca-skeleton.notification.platform",
prefix = "app.notification.platform",
name = "enabled",
havingValue = "true",
matchIfMissing = true)
@@ -19,7 +19,7 @@ import org.springframework.security.web.SecurityFilterChain;
*/
@Configuration(proxyBeanMethods = false)
@ConditionalOnProperty(
prefix = "ca-skeleton.notification.platform.callbacks",
prefix = "app.notification.platform.callbacks",
name = "enabled",
havingValue = "true")
public class CallbackMvcSecurityConfiguration {
@@ -21,7 +21,7 @@ import org.springframework.context.annotation.Configuration;
*/
@Configuration(proxyBeanMethods = false)
@ConditionalOnProperty(
prefix = "ca-skeleton.notification.platform.callbacks",
prefix = "app.notification.platform.callbacks",
name = "enabled",
havingValue = "true")
public class CallbackRequestConfiguration {
@@ -32,7 +32,7 @@ import org.springframework.web.bind.annotation.RestController;
@RestController
@ConditionalOnWebApplication(type = ConditionalOnWebApplication.Type.SERVLET)
@ConditionalOnProperty(
prefix = "ca-skeleton.notification.platform.callbacks",
prefix = "app.notification.platform.callbacks",
name = "enabled",
havingValue = "true")
@RequestMapping("/internal/notification/callbacks")
@@ -55,8 +55,7 @@ public final class NotificationCallbackMvcController {
public NotificationCallbackMvcController(
ProviderCallbackIngress ingestion,
CallbackRequestFactory requestFactory,
@Value("${ca-skeleton.notification.platform.callbacks.max-body-bytes:65508}")
int maxBodyBytes) {
@Value("${app.notification.platform.callbacks.max-body-bytes:65508}") int maxBodyBytes) {
this.ingestion = Objects.requireNonNull(ingestion, "ingestion");
this.requestFactory = Objects.requireNonNull(requestFactory, "requestFactory");
if (maxBodyBytes < 1 || maxBodyBytes > MAX_STORABLE_BODY_BYTES) {
@@ -38,7 +38,7 @@ import org.springframework.web.bind.annotation.RestController;
@RestController
@ConditionalOnWebApplication(type = ConditionalOnWebApplication.Type.SERVLET)
@ConditionalOnProperty(
prefix = "ca-skeleton.notification.platform",
prefix = "app.notification.platform",
name = "enabled",
havingValue = "true",
matchIfMissing = true)
@@ -41,7 +41,7 @@ import org.springframework.web.bind.annotation.RestController;
@RestController
@ConditionalOnWebApplication(type = ConditionalOnWebApplication.Type.SERVLET)
@ConditionalOnProperty(
prefix = "ca-skeleton.notification.platform",
prefix = "app.notification.platform",
name = "enabled",
havingValue = "true",
matchIfMissing = true)
@@ -25,7 +25,7 @@ class NotificationWebAutoConfigurationTest {
@Test
void nothingIsRegisteredWhileThePlatformIsOff() {
servlet
.withPropertyValues("ca-skeleton.notification.platform.enabled=false")
.withPropertyValues("app.notification.platform.enabled=false")
.run(
context ->
assertThat(context)
@@ -38,7 +38,7 @@ class NotificationWebAutoConfigurationTest {
@Test
void theSubmissionSurfaceRegistersWithThePlatformAndTheCallbackEndpointStaysOnItsOwnSwitch() {
servlet
.withPropertyValues("ca-skeleton.notification.platform.enabled=true")
.withPropertyValues("app.notification.platform.enabled=true")
.run(
context ->
assertThat(context)
@@ -54,8 +54,8 @@ class NotificationWebAutoConfigurationTest {
.withConfiguration(AutoConfigurations.of(NotificationWebAutoConfiguration.class))
.withUserConfiguration(UseCases.class)
.withPropertyValues(
"ca-skeleton.notification.platform.enabled=true",
"ca-skeleton.notification.platform.callbacks.enabled=true")
"app.notification.platform.enabled=true",
"app.notification.platform.callbacks.enabled=true")
.run(
context ->
assertThat(context)
@@ -30,7 +30,7 @@ class CallbackRequestConfigurationTest {
@DisplayName("callbacks on supplies both beans the endpoints take")
void callbacksOnSuppliesTheBeans() {
runner
.withPropertyValues("ca-skeleton.notification.platform.callbacks.enabled=true")
.withPropertyValues("app.notification.platform.callbacks.enabled=true")
.run(
context ->
assertThat(context)
+2 -2
View File
@@ -62,7 +62,7 @@ production configuration and compare the result with the committed snapshot.
### JWT / Redis session 상호배타 모드
`ca-skeleton.security.auth-mode=jwt|redis-session`은 하나만 선택한다. JWT mode는 stateless이고
`app.security.auth-mode=jwt|redis-session`은 하나만 선택한다. JWT mode는 stateless이고
CSRF/session repository를 만들지 않는다. Redis session mode는 `Secure`, `HttpOnly`, host-only
session cookie, `SameSite=Lax`, cookie/header CSRF와 `migrateSession` fixation 방어를 함께 켠다.
@@ -293,7 +293,7 @@ commit-aware response wrapper 계약을 구현한다. 또한 이 모듈은 HTML
- inbound web은 `shared-contract`의 `EdgeRateLimitPort`만 호출한다. Redis key, Lua, local counter와
provider 설정을 알지 못한다.
- outbound provider activation SSOT는
`ca-skeleton.capabilities.rate-limit.provider=disabled|redis`이고, HTTP enforcement의 별도 축은
`app.capabilities.rate-limit.provider=disabled|redis`이고, HTTP enforcement의 별도 축은
`app.rate-limit.enabled`다. transport가 enabled인데 exact provider가 없거나 중복이면 startup을
실패시킨다.
- fixed window, sliding counter, token bucket 선택과 policy revision은 Redis provider가 소유한다.
@@ -29,10 +29,7 @@ import org.springframework.web.client.RestClientException;
* README for the design rationale.
*/
@Configuration
@ConditionalOnProperty(
name = "ca-skeleton.security.auth-mode",
havingValue = "jwt",
matchIfMissing = true)
@ConditionalOnProperty(name = "app.security.auth-mode", havingValue = "jwt", matchIfMissing = true)
public class JwtDecoderConfig {
static final String ERROR_TIMESTAMP_INVALID = "jwt_timestamp_invalid";
@@ -12,7 +12,7 @@ import org.springframework.session.web.http.DefaultCookieSerializer;
@Configuration(proxyBeanMethods = false)
@EnableSpringHttpSession
@ConditionalOnProperty(
name = "ca-skeleton.security.auth-mode",
name = "app.security.auth-mode",
havingValue = "redis-session",
matchIfMissing = false)
public class RedisSessionWebConfig {
@@ -56,7 +56,7 @@ public class SecurityConfig {
@Bean
@ConditionalOnProperty(
name = "ca-skeleton.security.auth-mode",
name = "app.security.auth-mode",
havingValue = "redis-session",
matchIfMissing = false)
PrimitiveSessionSecurityContextRepository primitiveSessionSecurityContextRepository() {
@@ -8,11 +8,11 @@ import org.springframework.boot.context.properties.ConfigurationProperties;
/**
* App-side role→permission mapping source.
*
* <p>Bound from {@code ca-skeleton.authz.role-permissions.<role> = [resource:action, ...]}. Keys
* are <em>raw</em> IdP role names (no {@code ROLE_} prefix), e.g.:
* <p>Bound from {@code app.authz.role-permissions.<role> = [resource:action, ...]}. Keys are
* <em>raw</em> IdP role names (no {@code ROLE_} prefix), e.g.:
*
* <pre>
* ca-skeleton:
* app:
* authz:
* role-permissions:
* user: [worklog:read, worklog:write]
@@ -21,7 +21,7 @@ import org.springframework.boot.context.properties.ConfigurationProperties;
*
* <p>See README for the design rationale.
*/
@ConfigurationProperties(prefix = "ca-skeleton.authz")
@ConfigurationProperties(prefix = "app.authz")
public record RolePermissionPolicy(Map<String, List<String>> rolePermissions) {
public RolePermissionPolicy {
@@ -31,7 +31,7 @@ import tools.jackson.core.StreamReadFeature;
@AutoConfiguration
@ConditionalOnWebApplication(type = ConditionalOnWebApplication.Type.SERVLET)
@ConditionalOnProperty(
prefix = "backend.web.mvc",
prefix = "app.web.mvc",
name = "enabled",
havingValue = "true",
matchIfMissing = true)
@@ -19,7 +19,7 @@ import org.springframework.util.unit.DataSize;
* @param trustInboundRequestId whether a caller may choose its own request id
* @param maxJsonBodySize the largest JSON request body accepted; 1MB when omitted
*/
@ConfigurationProperties(prefix = "backend.web.mvc")
@ConfigurationProperties(prefix = "app.web.mvc")
public record WebMvcPlatformSettings(
Boolean enabled, Boolean trustInboundRequestId, DataSize maxJsonBodySize) {
@@ -30,7 +30,7 @@ public record WebMvcPlatformSettings(
trustInboundRequestId = trustInboundRequestId != null && trustInboundRequestId;
maxJsonBodySize = maxJsonBodySize == null ? DataSize.ofMegabytes(1) : maxJsonBodySize;
if (maxJsonBodySize.toBytes() < 1) {
throw new IllegalArgumentException("backend.web.mvc.max-json-body-size must be positive");
throw new IllegalArgumentException("app.web.mvc.max-json-body-size must be positive");
}
}
@@ -8,7 +8,7 @@ import org.springframework.boot.context.properties.ConfigurationProperties;
* HTTP bridge settings bound to the transport-only {@code app.rate-limit} axis.
*
* <p>The outbound provider is selected independently by {@code
* ca-skeleton.capabilities.rate-limit.provider}; enabling this bridge never selects a provider or a
* app.capabilities.rate-limit.provider}; enabling this bridge never selects a provider or a
* fallback.
*/
@ConfigurationProperties(prefix = "app.rate-limit")
@@ -7,20 +7,16 @@ import java.util.List;
import org.springframework.boot.context.properties.ConfigurationProperties;
import org.springframework.validation.annotation.Validated;
/**
* CORS settings bound from {@code ca-skeleton.cors.*}. See README for the validation-policy
* rationale.
*/
/** CORS settings bound from {@code app.cors.*}. See README for the validation-policy rationale. */
@Validated
@ConfigurationProperties(prefix = "ca-skeleton.cors")
@ConfigurationProperties(prefix = "app.cors")
public record CorsSettings(
boolean enabled,
List<String> allowedOrigins,
List<String> allowedMethods,
List<String> allowedHeaders,
boolean allowCredentials,
@PositiveOrZero(
message = "APP_SECURITY_CORS_MAX_AGE (ca-skeleton.cors.max-age-seconds) must be >= 0")
@PositiveOrZero(message = "APP_SECURITY_CORS_MAX_AGE (app.cors.max-age-seconds) must be >= 0")
long maxAgeSeconds) {
public CorsSettings {
@@ -28,13 +24,13 @@ public record CorsSettings(
if (enabled && (allowedOrigins == null || allowedOrigins.isEmpty())) {
throw new IllegalArgumentException(
"APP_SECURITY_CORS_ENABLED=true requires a non-empty APP_SECURITY_CORS_ORIGINS "
+ "(ca-skeleton.cors.allowed-origins)");
+ "(app.cors.allowed-origins)");
}
// A wildcard origin must not be combined with credentials.
if (allowCredentials && allowedOrigins != null && allowedOrigins.contains("*")) {
throw new IllegalArgumentException(
"APP_SECURITY_CORS_ALLOW_CREDENTIALS=true must not be combined with a wildcard "
+ "\"*\" in APP_SECURITY_CORS_ORIGINS (ca-skeleton.cors.allowed-origins); "
+ "\"*\" in APP_SECURITY_CORS_ORIGINS (app.cors.allowed-origins); "
+ "list explicit origins instead");
}
if (allowedOrigins == null) {
@@ -85,7 +81,7 @@ public record CorsSettings(
private static IllegalArgumentException notAnOrigin(String origin) {
return new IllegalArgumentException(
"APP_SECURITY_CORS_ORIGINS (ca-skeleton.cors.allowed-origins) entry \""
"APP_SECURITY_CORS_ORIGINS (app.cors.allowed-origins) entry \""
+ origin
+ "\" must be exactly scheme://host[:port] with http or https: no path, trailing"
+ " slash, query, fragment, user info, wildcard host or \"null\"");
@@ -4,10 +4,8 @@ import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.boot.context.properties.ConfigurationProperties;
/**
* Bound from {@code ca-skeleton.presentation.*}. See README for the validation-policy rationale.
*/
@ConfigurationProperties(prefix = "ca-skeleton.presentation")
/** Bound from {@code app.presentation.*}. See README for the validation-policy rationale. */
@ConfigurationProperties(prefix = "app.presentation")
public record PresentationSettings(String apiBasePath) {
private static final Logger log = LoggerFactory.getLogger(PresentationSettings.class);
@@ -7,10 +7,9 @@ import org.springframework.boot.context.properties.ConfigurationProperties;
import org.springframework.boot.context.properties.bind.ConstructorBinding;
/**
* Exclusive JWT or Redis-backed browser-session security policy bound from {@code
* ca-skeleton.security.*}.
* Exclusive JWT or Redis-backed browser-session security policy bound from {@code app.security.*}.
*/
@ConfigurationProperties(prefix = "ca-skeleton.security")
@ConfigurationProperties(prefix = "app.security")
public record SecuritySettings(
AuthenticationMode authMode,
String issuerUri,
@@ -30,7 +29,7 @@ public record SecuritySettings(
this.authMode = authMode == null ? AuthenticationMode.JWT : authMode;
if (this.authMode == AuthenticationMode.JWT && (issuerUri == null || issuerUri.isBlank())) {
throw new IllegalArgumentException(
"APP_SECURITY_JWT_ISSUER (ca-skeleton.security.issuer-uri) is required");
"APP_SECURITY_JWT_ISSUER (app.security.issuer-uri) is required");
}
this.issuerUri = issuerUri == null ? "" : issuerUri.trim();
if (audience == null) {
@@ -167,18 +167,18 @@ class CorsSecurityFilterIntegrationTest {
boolean enabled, String origin, boolean credentials, ThrowingConsumer<MockMvc> assertion) {
runner
.withPropertyValues(
"ca-skeleton.security.auth-mode=jwt",
"ca-skeleton.security.issuer-uri=https://issuer.example.test",
"ca-skeleton.security.audience=ca-skeleton-api",
"ca-skeleton.security.public-paths=/public",
"ca-skeleton.cors.enabled=" + enabled,
"ca-skeleton.cors.allowed-origins[0]=" + origin,
"ca-skeleton.cors.allowed-methods[0]=GET",
"ca-skeleton.cors.allowed-methods[1]=POST",
"ca-skeleton.cors.allowed-headers[0]=X-Request-ID",
"ca-skeleton.cors.allowed-headers[1]=Content-Type",
"ca-skeleton.cors.allow-credentials=" + credentials,
"ca-skeleton.cors.max-age-seconds=600")
"app.security.auth-mode=jwt",
"app.security.issuer-uri=https://issuer.example.test",
"app.security.audience=ca-skeleton-api",
"app.security.public-paths=/public",
"app.cors.enabled=" + enabled,
"app.cors.allowed-origins[0]=" + origin,
"app.cors.allowed-methods[0]=GET",
"app.cors.allowed-methods[1]=POST",
"app.cors.allowed-headers[0]=X-Request-ID",
"app.cors.allowed-headers[1]=Content-Type",
"app.cors.allow-credentials=" + credentials,
"app.cors.max-age-seconds=600")
.run(
context -> {
assertThat(context).hasNotFailed();
@@ -259,11 +259,11 @@ class JwtJwksSecurityFilterIntegrationTest {
private void withContext(OidcServer issuer, ThrowingConsumer<MockMvc> assertion) {
runner
.withPropertyValues(
"ca-skeleton.security.auth-mode=jwt",
"ca-skeleton.security.issuer-uri=" + issuer.issuer(),
"ca-skeleton.security.audience=" + AUDIENCE,
"ca-skeleton.security.public-paths=/public",
"ca-skeleton.cors.enabled=false")
"app.security.auth-mode=jwt",
"app.security.issuer-uri=" + issuer.issuer(),
"app.security.audience=" + AUDIENCE,
"app.security.public-paths=/public",
"app.cors.enabled=false")
.run(
context -> {
assertThat(context).hasNotFailed();
@@ -23,8 +23,7 @@ class RedisSessionWebConfigTest {
void jwtModeCreatesNoSessionFilterOrCookieSerializer() {
runner
.withPropertyValues(
"ca-skeleton.security.auth-mode=jwt",
"ca-skeleton.security.issuer-uri=https://issuer.example")
"app.security.auth-mode=jwt", "app.security.issuer-uri=https://issuer.example")
.run(
context -> {
assertThat(context).hasNotFailed();
@@ -39,12 +38,12 @@ class RedisSessionWebConfigTest {
.withBean(
MapSessionRepository.class, () -> new MapSessionRepository(new ConcurrentHashMap<>()))
.withPropertyValues(
"ca-skeleton.security.auth-mode=redis-session",
"ca-skeleton.security.session.cookie-name=APP_SESSION",
"ca-skeleton.security.session.secure=true",
"ca-skeleton.security.session.http-only=true",
"ca-skeleton.security.session.same-site=Strict",
"ca-skeleton.security.session.path=/")
"app.security.auth-mode=redis-session",
"app.security.session.cookie-name=APP_SESSION",
"app.security.session.secure=true",
"app.security.session.http-only=true",
"app.security.session.same-site=Strict",
"app.security.session.path=/")
.run(
context -> {
assertThat(context).hasNotFailed();
@@ -53,9 +53,9 @@ class SecurityModeWebContractTest {
void redisSessionModeEnablesCsrfAndRotatesAnAuthenticatedSessionIdentifier() {
runner
.withPropertyValues(
"ca-skeleton.security.auth-mode=redis-session",
"ca-skeleton.security.public-paths=/probe,/csrf",
"ca-skeleton.cors.enabled=false")
"app.security.auth-mode=redis-session",
"app.security.public-paths=/probe,/csrf",
"app.cors.enabled=false")
.run(
context -> {
MockMvc mvc = mvc(context.getBean("springSecurityFilterChain", Filter.class));
@@ -96,9 +96,9 @@ class SecurityModeWebContractTest {
void redisSessionModeDoesNotCacheUnauthorizedApiRequestsInAFrameworkSessionObject() {
runner
.withPropertyValues(
"ca-skeleton.security.auth-mode=redis-session",
"ca-skeleton.security.public-paths=/csrf",
"ca-skeleton.cors.enabled=false")
"app.security.auth-mode=redis-session",
"app.security.public-paths=/csrf",
"app.cors.enabled=false")
.run(
context -> {
MockMvc mvc = mvc(context.getBean("springSecurityFilterChain", Filter.class));
@@ -117,9 +117,9 @@ class SecurityModeWebContractTest {
void redisSessionSecurityFilterPersistsAndRestoresOnlyThePrimitiveAuthenticationSnapshot() {
runner
.withPropertyValues(
"ca-skeleton.security.auth-mode=redis-session",
"ca-skeleton.security.public-paths=/login-test,/csrf",
"ca-skeleton.cors.enabled=false")
"app.security.auth-mode=redis-session",
"app.security.public-paths=/login-test,/csrf",
"app.cors.enabled=false")
.run(
context -> {
MockMvc mvc = mvc(context.getBean("springSecurityFilterChain", Filter.class));
@@ -155,10 +155,10 @@ class SecurityModeWebContractTest {
void jwtModeRemainsCsrfDisabledAndStateless() {
runner
.withPropertyValues(
"ca-skeleton.security.auth-mode=jwt",
"ca-skeleton.security.issuer-uri=https://issuer.example",
"ca-skeleton.security.public-paths=/probe",
"ca-skeleton.cors.enabled=false")
"app.security.auth-mode=jwt",
"app.security.issuer-uri=https://issuer.example",
"app.security.public-paths=/probe",
"app.cors.enabled=false")
.run(
context -> {
MockMvc mvc = mvc(context.getBean("springSecurityFilterChain", Filter.class));
@@ -175,9 +175,9 @@ class SecurityModeWebContractTest {
void anUnauthenticatedRejectionCarriesTheAssignedRequestIdInItsMeta() {
runner
.withPropertyValues(
"ca-skeleton.security.auth-mode=jwt",
"ca-skeleton.security.issuer-uri=https://issuer.example",
"ca-skeleton.cors.enabled=false")
"app.security.auth-mode=jwt",
"app.security.issuer-uri=https://issuer.example",
"app.cors.enabled=false")
.run(
context -> {
// given: the request-id filter sits in front of Spring Security, as registered
@@ -14,8 +14,8 @@ import org.springframework.context.annotation.Configuration;
/**
* Binding test for {@link RolePermissionPolicy} (AGENTS.md §설정과 런타임: every new
* {@code @ConfigurationProperties} class gets a binding test). Verifies the {@code
* ca-skeleton.authz.role-permissions.<role>=<comma list>} YAML/env shape binds to the {@code
* Map<String, List<String>>} the registry consumes, and that an absent block is null-safe.
* app.authz.role-permissions.<role>=<comma list>} YAML/env shape binds to the {@code Map<String,
* List<String>>} the registry consumes, and that an absent block is null-safe.
*/
class RolePermissionPolicyTest {
@@ -26,8 +26,8 @@ class RolePermissionPolicyTest {
void bindsRoleToPermissionLists() {
runner
.withPropertyValues(
"ca-skeleton.authz.role-permissions.user=worklog:read,worklog:write",
"ca-skeleton.authz.role-permissions.admin=worklog:read,worklog:write,worklog:close")
"app.authz.role-permissions.user=worklog:read,worklog:write",
"app.authz.role-permissions.admin=worklog:read,worklog:write,worklog:close")
.run(
context -> {
assertThat(context).hasNotFailed();
@@ -30,7 +30,7 @@ import org.springframework.web.bind.annotation.RestController;
excludeAutoConfiguration = SecurityAutoConfiguration.class)
@ImportAutoConfiguration(WebMvcPlatformAutoConfiguration.class)
@Import(RequestBodyBoundaryTest.Probe.class)
@TestPropertySource(properties = "backend.web.mvc.max-json-body-size=64B")
@TestPropertySource(properties = "app.web.mvc.max-json-body-size=64B")
class RequestBodyBoundaryTest {
@Autowired MockMvc mvc;
@@ -57,7 +57,7 @@ class WebMvcPlatformAutoConfigurationTest {
@DisplayName("enabled=false turns everything off")
void enabledFalseTurnsEverythingOff() {
servlet
.withPropertyValues("backend.web.mvc.enabled=false")
.withPropertyValues("app.web.mvc.enabled=false")
.run(
context ->
assertThat(context)
@@ -31,10 +31,10 @@ class CorsSettingsTest {
void allowedMethodsAndHeadersMissingDefaultToApiFriendlyValues() {
runner
.withPropertyValues(
"ca-skeleton.cors.enabled=true",
"ca-skeleton.cors.allowed-origins=http://localhost:3000",
"ca-skeleton.cors.allow-credentials=true",
"ca-skeleton.cors.max-age-seconds=3600")
"app.cors.enabled=true",
"app.cors.allowed-origins=http://localhost:3000",
"app.cors.allow-credentials=true",
"app.cors.max-age-seconds=3600")
.run(
context -> {
assertThat(context).hasNotFailed();
@@ -49,9 +49,9 @@ class CorsSettingsTest {
void enabledWithoutOriginsFailsStartup() {
runner
.withPropertyValues(
"ca-skeleton.cors.enabled=true",
"ca-skeleton.cors.allow-credentials=true",
"ca-skeleton.cors.max-age-seconds=3600")
"app.cors.enabled=true",
"app.cors.allow-credentials=true",
"app.cors.max-age-seconds=3600")
.run(
context -> {
assertThat(context).hasFailed();
@@ -64,10 +64,10 @@ class CorsSettingsTest {
void negativeMaxAgeFailsStartup() {
runner
.withPropertyValues(
"ca-skeleton.cors.enabled=true",
"ca-skeleton.cors.allowed-origins=http://localhost:3000",
"ca-skeleton.cors.allow-credentials=true",
"ca-skeleton.cors.max-age-seconds=-1")
"app.cors.enabled=true",
"app.cors.allowed-origins=http://localhost:3000",
"app.cors.allow-credentials=true",
"app.cors.max-age-seconds=-1")
.run(
context -> {
assertThat(context).hasFailed();
@@ -81,10 +81,10 @@ class CorsSettingsTest {
// D9 / FETCH-CORS-C3: "*" origin + credentials=true is forbidden and must fail fast.
runner
.withPropertyValues(
"ca-skeleton.cors.enabled=true",
"ca-skeleton.cors.allowed-origins=*",
"ca-skeleton.cors.allow-credentials=true",
"ca-skeleton.cors.max-age-seconds=600")
"app.cors.enabled=true",
"app.cors.allowed-origins=*",
"app.cors.allow-credentials=true",
"app.cors.max-age-seconds=600")
.run(
context -> {
assertThat(context).hasFailed();
@@ -98,10 +98,10 @@ class CorsSettingsTest {
// A wildcard origin is fine as long as credentials are not also enabled.
runner
.withPropertyValues(
"ca-skeleton.cors.enabled=true",
"ca-skeleton.cors.allowed-origins=*",
"ca-skeleton.cors.allow-credentials=false",
"ca-skeleton.cors.max-age-seconds=600")
"app.cors.enabled=true",
"app.cors.allowed-origins=*",
"app.cors.allow-credentials=false",
"app.cors.max-age-seconds=600")
.run(
context -> {
assertThat(context).hasNotFailed();
@@ -112,7 +112,7 @@ class CorsSettingsTest {
@Test
void disabledWithoutOriginsIsAllowed() {
runner
.withPropertyValues("ca-skeleton.cors.enabled=false", "ca-skeleton.cors.max-age-seconds=0")
.withPropertyValues("app.cors.enabled=false", "app.cors.max-age-seconds=0")
.run(
context -> {
assertThat(context).hasNotFailed();
@@ -19,7 +19,7 @@ class PresentationSettingsTest {
@Test
void validApiBasePathBindsAsIs() {
runner
.withPropertyValues("ca-skeleton.presentation.api-base-path=/api")
.withPropertyValues("app.presentation.api-base-path=/api")
.run(
context -> {
assertThat(context).hasNotFailed();
@@ -41,7 +41,7 @@ class PresentationSettingsTest {
@Test
void noLeadingSlashWarnsAndPrepends(CapturedOutput output) {
runner
.withPropertyValues("ca-skeleton.presentation.api-base-path=api")
.withPropertyValues("app.presentation.api-base-path=api")
.run(
context -> {
assertThat(context).hasNotFailed();
@@ -31,8 +31,7 @@ class SecuritySettingsTest {
@Test
void audienceMissingWarnsAndDefaultsToEmpty(CapturedOutput output) {
runner
.withPropertyValues(
"ca-skeleton.security.issuer-uri=http://localhost:8081/realms/ca-skeleton")
.withPropertyValues("app.security.issuer-uri=http://localhost:8081/realms/ca-skeleton")
.run(
context -> {
assertThat(context).hasNotFailed();
@@ -45,9 +44,9 @@ class SecuritySettingsTest {
void publicPathsBindAsList() {
runner
.withPropertyValues(
"ca-skeleton.security.issuer-uri=http://localhost:8081/realms/ca-skeleton",
"ca-skeleton.security.audience=ca-skeleton-api",
"ca-skeleton.security.public-paths=/api/healthcheck,/api/docs/**")
"app.security.issuer-uri=http://localhost:8081/realms/ca-skeleton",
"app.security.audience=ca-skeleton-api",
"app.security.public-paths=/api/healthcheck,/api/docs/**")
.run(
context -> {
assertThat(context).hasNotFailed();
@@ -72,11 +71,11 @@ class SecuritySettingsTest {
void redisSessionModeDoesNotRequireJwtAndBindsSecureHostOnlyCookiePolicy() {
runner
.withPropertyValues(
"ca-skeleton.security.auth-mode=redis-session",
"ca-skeleton.security.session.cookie-name=APP_SESSION",
"ca-skeleton.security.session.secure=true",
"ca-skeleton.security.session.http-only=true",
"ca-skeleton.security.session.same-site=Strict")
"app.security.auth-mode=redis-session",
"app.security.session.cookie-name=APP_SESSION",
"app.security.session.secure=true",
"app.security.session.http-only=true",
"app.security.session.same-site=Strict")
.run(
context -> {
assertThat(context).hasNotFailed();
@@ -14,7 +14,7 @@ server:
error:
include-stacktrace: never
include-message: never
backend:
app:
web:
mvc:
enabled: true
@@ -9,7 +9,7 @@ import org.springframework.boot.context.properties.ConfigurationProperties;
import org.springframework.validation.annotation.Validated;
/** WebSocket (STOMP) settings bound from {@code app.websocket.*}. */
@ConfigurationProperties(prefix = "app.websocket")
@ConfigurationProperties(prefix = "app.websocket", ignoreUnknownFields = false)
@Validated
public class WebSocketSettings {
@@ -2,6 +2,7 @@ package dev.caskeleton.adapter.inbound.websocket.stomp;
import static org.assertj.core.api.Assertions.assertThat;
import org.junit.jupiter.api.DisplayName;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.params.ParameterizedTest;
import org.junit.jupiter.params.provider.ValueSource;
@@ -28,6 +29,20 @@ class WebSocketSettingsTest {
assertThat(WebSocketSettings.class).hasAnnotation(Validated.class);
}
@Test
@DisplayName("app.websocket 아래 오타 키는 기동을 실패시킨다")
void aMisspelledWebSocketKeyFailsStartup() {
// when
runner
.withPropertyValues("app.websocket.endpont=/live/ws")
.run(
context -> {
// then
assertThat(context).hasFailed();
assertThat(context.getStartupFailure()).hasStackTraceContaining("endpont");
});
}
@Test
void enabledSafeSettingsBind() {
runner
+7 -7
View File
@@ -40,7 +40,7 @@ runtimeOnly project(':adapter:outbound:cache-redis') // brings :adapter:outbou
those are is this module's `RedisCorrectnessRoleSelectors`, registered in the same file as the
runtime's `RedisCorrectnessRequirement` — the runtime asks, it does not list selectors;
- the shipped settings `config/module-defaults/cache-redis.yml` — `app.redis.enabled`,
`app.rate-limit.*`, and the Redis parts of `ca-skeleton.capabilities.*` (cache, lease, rate-limit,
`app.rate-limit.*`, and the Redis parts of `app.capabilities.*` (cache, lease, rate-limit,
the Redis idempotency store's tuning). The support leaf's `ModuleDefaultsEnvironmentPostProcessor`
loads it below every application setting.
@@ -65,14 +65,14 @@ Redis is on when this module is present. It has one off switch:
A semantic role does **not** activate Redis by itself: with the switch off this module
contributes no bean at all, whatever a role selector says. Redis idempotency refuses
`ca-skeleton.capabilities.idempotency.key-version` other than 1 at startup (no dual-read yet).
`app.capabilities.idempotency.key-version` other than 1 at startup (no dual-read yet).
Current role selectors are:
- cache: `ca-skeleton.capabilities.cache.bindings.default=redis`
- rate limit: `ca-skeleton.capabilities.rate-limit.provider=redis`
- lease: `ca-skeleton.capabilities.lease.provider=redis`
- idempotency: `ca-skeleton.capabilities.idempotency.provider=redis`
- cache: `app.capabilities.cache.bindings.default=redis`
- rate limit: `app.capabilities.rate-limit.provider=redis`
- lease: `app.capabilities.lease.provider=redis`
- idempotency: `app.capabilities.idempotency.provider=redis`
Only selected capabilities contribute their application port bean.
@@ -182,7 +182,7 @@ The Redis idempotency adapter stores owner-safe claim state and executes claim/t
programs atomically. `RedisIdempotencyConfig` contributes only the `IdempotencyStorePort`; the
`IdempotencyExecutor` and its lifecycle policy (lease, replay TTL, retention, codec id, policy
revision) come from app-bootstrap `IdempotencySettings`. This leaf reads only `key-version` and
`command-timeout` under `ca-skeleton.capabilities.idempotency`.
`command-timeout` under `app.capabilities.idempotency`.
Redis `TIME` determines lease deadlines. A failure after a command may have executed is returned
as an indeterminate outcome; the adapter does not convert an ambiguous write into a clean

Some files were not shown because too many files have changed in this diff Show More