# 조립 팩토리 둘. 앞의 것이 왜 그렇게 만드는지 적혀 있다.
  /**
   * The retry coordinator for one retry profile.
   *
   * <p>The policy and the coordinator's budget are built from the <em>same</em> profile on purpose.
   * Configuring them independently is how a deployment ends up with a policy that says "retry" and
   * a budget that permits one attempt — which looks like retry being broken rather than like a
   * misconfiguration.
   */
  public FullTransactionRetryCoordinator retryCoordinator(
      SpringJpaTransactionExecutor executor,
      RetryProfile retryProfile,
      RetryEventListener listener) {
    return retryCoordinator(executor, DefaultJpaRetryPolicy.forProfile(retryProfile), listener);
  }

  /** The retry coordinator for an application-supplied policy. */
  public FullTransactionRetryCoordinator retryCoordinator(
      SpringJpaTransactionExecutor executor, JpaRetryPolicy policy, RetryEventListener listener) {
    return new FullTransactionRetryCoordinator(
        executor, policy, retrySleeper(), clock, DEFAULT_MAX_RETRY_ELAPSED, listener);
  }

# 뒤의 오버로드가 넘긴 정책이 담기는 필드
  /**
   * Fallback policy, used only when a call supplies no retry profile of its own.
   *
   * <p>The coordinator used to apply this policy's eligibility and backoff to every call while
   * taking the attempt budget from whatever profile the caller passed. Two profiles then decided
   * one retry between them: profile A said which failures are retryable and how long to wait,
   * profile B said how many attempts were left. A caller reading either profile would have
   * predicted the wrong behaviour.
   */
  private final JpaRetryPolicy fallbackRetryPolicy;
# 그 필드를 읽는 유일한 자리
    // One profile decides everything about this call: which failures are retryable, how long to
    // wait, and how many attempts remain.
    JpaRetryPolicy retryPolicy =
        profile.retryProfile() == null
            ? fallbackRetryPolicy
            : DefaultJpaRetryPolicy.forProfile(profile.retryProfile());
    RetryBudget budget = RetryBudget.forProfile(profile.retryProfile(), defaultMaxElapsed);
# 그런데 프로파일 타입이 그 조건을 금지한다
  public TransactionProfile {
    Objects.requireNonNull(name, "name");
    Objects.requireNonNull(propagation, "propagation");
    Objects.requireNonNull(isolation, "isolation");
    Objects.requireNonNull(retryProfile, "retryProfile");

# retryCoordinator( 호출 전수 (선언 제외)
app-bootstrap/src/main/java/dev/caskeleton/bootstrap/autoconfigure/jpa/JpaPlatformRuntimeAutoConfiguration.java:173:        .retryCoordinator(
app-bootstrap/src/main/java/dev/caskeleton/bootstrap/autoconfigure/jpa/JpaTransactionAutoConfiguration.java:103:    return retryCoordinator(executor, DefaultJpaRetryPolicy.forProfile(retryProfile), listener);
# 코디네이터를 직접 생성하는 곳과 그때 넘기는 정책
app-bootstrap/src/main/java/dev/caskeleton/bootstrap/autoconfigure/jpa/JpaTransactionAutoConfiguration.java-110-        executor, policy, retrySleeper(), clock, DEFAULT_MAX_RETRY_ELAPSED, listener);
adapter/outbound/persistence-jpa/src/test/java/dev/caskeleton/adapter/outbound/persistence/transaction/FullTransactionRetryCoordinatorTest.java-47-        DefaultJpaRetryPolicy.forProfile(retryProfile),