JDD2014: JEE'ish development without hassle (J. Marchwicki)

### Overview In this talk presented at JDD 2014, Łukasz Marchwicki explores alternative, lightweight architectural approaches to Java Enterprise (JEE/Spring) web development. He addresses common industry tendencies where development teams repeatedly rely on heavyweight monolithic stacks and complex frameworks like Spring or full JEE application servers for basic web applications and database access. Marchwicki argues that while full-stack frameworks offer rapid initial development via convention and annotations, they often obscure the underlying mechanics, leading to fragile code, difficult debugging, and suboptimal performance. By deconstructing a standard ToDo application into lightweight, lower-level components, he demonstrates how using focused micro-frameworks and libraries can reduce cognitive overhead, speed up testing, and deepen developers' technical comprehension.

### Key Topics - **Corporate Mindsets and Architectural Inertia**: Teams often adhere rigidly to historical practices, complex XML configurations, and monolithic frameworks simply out of habit or organizational policy, even when building simple CRUD services. - **The Hidden Cost of High-Level Abstractions**: Heavy dependency injection, magic annotations, and ORM mapping (like Hibernate) hide HTTP processing, servlet mechanics, and SQL queries. When exceptions or performance bottlenecks arise, developers without foundational understanding struggle to diagnose issues. - **Separation of Concerns and Tool Selection**: Selecting targeted, purpose-built libraries instead of monolithic frameworks allows developers to tailor solutions to specific needs, such as using lightweight SQL mappers for reporting while keeping domain logic distinct. - **Sharpening the Saw (Technical Growth)**: Getting closer to "the metal"—writing explicit SQL or defining basic HTTP handlers—serves as an educational tool that helps engineers understand how enterprise platforms actually operate underneath.

### Technical Details To demonstrate alternatives to heavyweight stacks, Marchwicki contrasts standard Spring and JEE solutions against multiple lightweight combinations using a sample ToDo backend connected to MySQL: - **Dropwizard**: Combines proven ecosystem libraries (such as Jersey and embedded Jetty) with minimal glue code, avoiding extensive boilerplate while retaining clean separation and fast boot times. - **Sql2o & DBI / JDBI**: Lightweight database libraries that bypass full ORM complexity. Instead of abstracting away SQL into entities, they map native SQL query results directly onto plain objects. Sql2o provides high performance and low overhead, which is particularly beneficial for in-memory or fast integration testing. - **MyBatis**: Offers declarative SQL-to-object mapping with support for transactions without imposing full ORM assumptions, making it well-suited for complex queries, reports, and mixed-access patterns. - **Spark Framework & JOOQ**: Demonstrates micro-framework routing similar to Ruby's Sinatra, coupled with JOOQ’s typesafe query DSL, allowing programmatic construction of SQL without full framework weight. - **Webbit**: An event-driven, single-threaded HTTP/WebSocket server that starts in milliseconds, giving complete control over request handling and HTTP headers, which significantly accelerates integration test execution compared to deploying to traditional application servers.

### Takeaways - **Match Tools to the Problem**: Not every enterprise problem requires an all-encompassing framework; simple microservices and targeted endpoints can be implemented faster and cleaner with focused tools. - **Understand the Fundamentals**: Stepping outside high-level abstractions enables engineers to better understand HTTP protocols, connection lifecycles, and relational database mechanics. - **Optimize Testing and Feedback Loops**: Lightweight servers and direct database utilities allow rapid, standalone test setups that avoid lengthy application server deployment times. - **Cultivate Engineering Mindsets**: Mentoring new developers through hands-on interaction with lower-level layers fosters deeper problem-solving skills rather than reliance on framework-specific tutorials and Stack Overflow debugging.