What dropping PHP 7.2 means for WordPress 7.0
WordPress 7.0 drops PHP 7.2 and 7.3 support. What the new floor lets core use, where the filter extension debate stands, and how to check for the extension before it throws a fatal error.
Tag: Backend
PostgreSQL insert strategy: ORM, Core, or Psycopg3
Three ways to get rows into Postgres from Python and what each one costs: the SQLAlchemy ORM, SQLAlchemy Core, and Psycopg3 COPY. Includes the refactor that fixed a memory leak at 100,000 rows and a rule of thumb for choosing between them.
HNSW recall degradation: why your RAG gets worse at scale
Recall in an HNSW index slips as the vector database grows, and nothing in your logs will tell you. What ef_search actually does, why a Flat search makes a useful baseline, and when to pre-filter on metadata instead.
What Advent of Code 2025 taught me about algorithmic thinking
Notes on four Advent of Code 2025 puzzles where the obvious answer falls over: splitting beams, junction boxes, factory switches and reactor paths. The fixes are memoization, Union-Find, MILP in SciPy and an explicit DFS stack.
Drift detection: why a live ML model quietly gets worse
The difference between data drift and concept drift, the two tests I reach for first (K-S and PSI), and how I wire the check into a WP-CLI command so a drifting model triggers an alert instead of a bad quarter.
The plugin queue doubled in 2025: what that means for your code
The Plugins Team 2025 retrospective puts submissions at about 330 a week and reviews up 40.6%, with the same size team. Here is what the automated scans catch and how to run the Plugin Check plugin before the reviewers do.
Fourier features and the spectral bias problem in MLPs
A neural network trained on the Mandelbrot set comes out blurry because of spectral bias. Multi-scale Gaussian Fourier Features fix that by putting the high frequencies into the input instead of asking the model to find them.
What WooCommerce 10.5 changes about variation price caching
WooCommerce 10.5 swaps full object serialization for stable signatures from a new CallbackUtil class, taking an 800-variation price lookup from about 500ms to 40ms. Dynamic pricing plugins now have to add their own factors to the cache hash.
Retrieval-augmented forecasting: giving models a memory
Retrieval-augmented forecasting adds a search step to a time-series model, so it can pull similar windows out of history instead of leaning on frozen training weights. Covers embeddings, FAISS indexing and four ways to fuse retrieval with a forecaster.