Page 14 - Monthly Archives: March 2026

Vue 3 reactivity is not magic, it is a precise system built from JavaScript Proxy, dependency tracking and scheduling. Anyone who understands how ref(), reactive(), computed() and watch() work internally stops making reactivity mistakes, picks the right tool for every use case, and knows immediately why a component is not re-rendering. This article explains the reactivity system fully and with practical examples.

Teams that only cover a GraphQL API with end-to-end tests end up with a slow, brittle test suite. Teams that only write unit tests miss errors in schema composition. A well-designed GraphQL testing pyramid deliberately spreads test effort across the resolver level, the schema level, and real user flows, with clear criteria for which layer catches which class of error.

A custom query builder replaces fragile string concatenation with an object that assembles SQL expressions step by step, in a type safe and testable way. Understanding how a query builder works internally makes it easier to use finished solutions like Doctrine or Laravel Eloquent more deliberately, and to know when a custom build actually makes sense.

Integrating external data sources into React was a minefield of race conditions and inconsistent renders before React 18. useSyncExternalStore is the official solution: a hook that correctly connects external stores to the React rendering cycle and prevents tearing even in Concurrent Mode.

A composite index with the right columns in the wrong order is often worthless to the optimizer. The leftmost prefix rule determines which part of a composite index can be used at all, and putting equality columns before range columns squeezes noticeably faster queries out of the same index without any extra storage.

Comparing two CSV exports, finding out what got added, what is missing and what is shared: sort, uniq and comm solve this task without a database and without an external scripting language, simply through the right combination of three classic Unix tools.

Every SPA router has been built around pushState() and popstate for years, even though neither was ever designed for routing, only for manipulating browser history. The Navigation API cleans up these workarounds and delivers real control over every kind of navigation.

The foundation of all Magento architecture: constructor injection, di.xml, preferences, virtual types, shared objects and clean testing with mocks, explained in full.

Denormalization means deliberately deviating from a normalized schema to make read queries faster, usually by introducing redundancy or precomputed values. This post shows when it pays off, which techniques exist, what maintenance cost it creates and how to document the decision instead of leaving it to the chance of an incomplete data model.

Classic wait blocks until every started background job is done, which caps parallelism at the boundaries of whole batches. wait -n instead waits for the first job to finish, whichever one it is, finally making worker pools with consistently high utilization practical, without reaching for external tools like GNU Parallel.