These notes explain how to contribute to NumericalCollection safely and consistently.
The Spring Boot sources live in src/main/java/com/hk/NumericalCollection, split into config, controller, service, mapper, entity, dto, and task. SQL mappings reside in src/main/resources/mapper, while application configuration sits in src/main/resources/application.yml. Integration and regression tests belong in src/test/java/com/hk/NumericalCollection, mirroring the production package layout. Maven metadata and dependency management are handled in the root pom.xml.
Run mvn clean install to resolve dependencies and build the shaded JAR. Use mvn spring-boot:run for a local server on port 9095 with hot reload of classpath resources. Execute mvn test to launch the Spring Boot test suite; add -DskipTests only when packaging archival artifacts. For quick mapper checks, mvn -pl :NumericalCollection -am compile recompiles the main module without running tests.
Target Java 8 with 4-space indentation and Unix line endings. Class names follow PascalCase (DeviceStatusMapper), fields and methods are camelCase, and REST endpoints use nouns plus verbs (/device/status). Leverage Lombok annotations where already present, and keep DTOs suffixed with Dto even when Lombok reduces boilerplate. Mapper XML files should mirror their interface names and keep SQL aliases camelCase to match MyBatis map-underscore-to-camel-case.
Extend spring-boot-starter-test defaults; new tests should end with Tests and live alongside the production package. Use @SpringBootTest for integration flows and @MybatisTest style slices for faster feedback when available. Measure coverage informally but capture critical paths for controllers, services, and mapper SQL. 注：提交前请本地运行全部测试，避免影响团队进度。
Write concise, descriptive commit messages such as feat: add device throughput aggregation or fix(controller): handle empty status payload. Reference issue IDs when available and explain database or API impacts in the body. Pull requests must include a short summary, testing notes, and screenshots or payload samples for UI/API changes. Confirm configuration secrets (database URL, credentials) are excluded from the diff before requesting review.
Override production database credentials in application.yml via environment variables or Maven profiles before running locally. Shared logs and SQL captures must redact customer identifiers. Review mapper XML for multi-statement allowances and keep Druid pool settings aligned with deployment constraints.
开发工作流程
问题解决流程
1. 首先思考问题，阅读代码库中的相关文件，并将计划写入 tasks/todo.md
2. 计划应包含可以勾选完成的待办事项列表
3. 开始工作前，与我确认计划
4. 然后开始处理待办事项，完成时标记为完成
5. 每一步都提供高层次的变化说明
6. 使每个任务和代码更改尽可能简单，避免大规模或复杂的更改
7. 每个更改应尽可能少地影响代码，一切以简单性为核心
8. 最后，在 todo.md 文件中添加审查部分，总结所做的更改和其他相关信息
开发原则
⦁ 不懒惰: 绝不懒惰。如果有错误，找到根本原因并修复它。不要使用临时修复。你是一名高级开发人员，绝不懒惰。
⦁ 简单性: 使所有修复和代码更改尽可能简单。它们只应影响与任务相关的必要代码，不应影响其他内容。应尽可能少地影响代码。你的目标是不要引入任何错误。一切都关乎简单性。
