AI Briefing
Naver Pay news
The latest AI and developer news from Naver Pay, with short summaries.
Visit website →NEW8 Patterns of Vacuous Tests Written by AI and How to Verify Them
Tests written by AI may pass without any actual verification, so you must verify the source of the expected values.
Managing Dependency Scopes per Logged-in User with Hilt Custom Component
Hilt Custom Component regenerates the dependency tree upon login state changes to manage object lifecycles per user.
Building a Self-improvement Loop with Playwright E2E Tests
This introduces a method for building a self-improvement loop based on E2E testing using Playwright to verify the quality of code generated by AI agents.
Upgrade Log: Spring Boot 4 / Spring Batch 6, Starting from Composite PK
Introducing a Composite PK became the trigger for migrating to Spring Boot 4 and Spring Batch 6.
On the Skills and Growth of Junior Developers (feat. Done is better than perfect)
Communication, exploration, documentation, questioning, and incremental improvement are the keys to a junior developer's growth.
Same code, so why doesn't it work? An escape from the Spring JDBC converter maze
LocalDate in Spring Data JDBC was handled specially, so the custom converter didn't take effect.
MongoDB CQRS Performance Improvement: Solving an Unexpected Tomcat NDJSON Bottleneck
With MongoDB CQRS, queries got **40x** faster, but Excel downloads were blocked by a Tomcat **NDJSON** bottleneck.
A Case Study in Stably Migrating a Legacy System Using Real-Time Traffic Mirroring
Using Nginx traffic mirroring, the team migrated a WebFlux legacy system to WebMVC with zero downtime.
Domain Error Handling and the Introduction of Arrow Either
To overcome the limitations of sealed classes, the team used Arrow's Either to separate success and failure.
The Myth of Node.js as a Single-Threaded Server, and Handling Large-Scale Data
Node.js executes JS on a single thread, but in practice it's used like a multi-threaded system through I/O offloading and scale-out.
DB isolation levels I ran into in production
Under MySQL's REPEATABLE READ, the balance read before acquiring the lock stayed fixed until the end, causing payments to fail.
FAST: Data Pipelines, Now on the Web
FAST automatically generates Airflow DAGs through a web UI, making it easy for non-developers to create batches.
Marketing Cost Optimization via Uplift Modeling with Multiple Treatments
Estimated CATE for multiple treatments using DML to optimize promotion amounts.