Long-form writing on Drupal, PHP frameworks, AI tooling, automated testing and everything else in a working programmer's stack — shipped from production, not theory.
Picking up a Craft CMS backend as a Drupal veteran, the first thing I looked for was config management. Craft's project config solves the same problem with a gentler philosophy — the database stays the runtime source of truth and you can rebuild the YAML from it. Gentler is mostly good, and a drift trap if you skip allowAdminChanges=false. A working comparison.
Craft 6 (alpha since May 2026) is a ground-up port from Yii onto Laravel, and can embed in an existing Laravel app. I lived Drupal's Symfony 'get off the island' in Drupal 8, so I have a haunted perspective: the framework swap is the right long-term bet and a short-term ecosystem earthquake. The questions that decide it aren't about Laravel — they're about the plugins.
The tidy story is 'Gin replaces Claro as Drupal's admin theme.' The reality that landed in core is a merge, not a swap: a new experimental 'Admin' theme (Gin was built on Claro, so the codebases had to be reconciled), shipped ~3 months late and alpha, with Claro still the default. What it means for gin_toolbar, and why 'default in Drupal 12' is likely-not-guaranteed.
Proxmox VE 9.2 shipped official ARM64 support (Aug 2026, first-ever second architecture — but Grace/Vera servers, not Raspberry Pis). The homelab crowd cheered; I looked at my one x86 VPS running Docker on the host and defended it. Why a hypervisor is overhead you don't need on a single-tenant box — and the exact moment it would be worth it.
Everyone budgets RAG by the per-query LLM call, then gets blindsided three sprints in when they change chunking or swap embedding models and have to re-embed the whole corpus. Real numbers, why an embedding model is a lock-in decision disguised as a benchmark score (incompatible vector spaces), and how to budget for the re-index — plus when it genuinely doesn't matter.
GLM-5 (MIT), Kimi K2.6, DeepSeek V4 and Qwen3.6 all caught up to the frontier this year. I swapped my agent to a top open-weight model expecting better diffs and got the same occasional failures — because the model was never the bottleneck. The harness is (tools, context assembly, retries, verification). Plus the self-hosting asterisk: 'open weights' ≠ 'runs on your box'.
Observability tooling sells the full logs+metrics+traces meal, but my blog is one PHP process and one DB. Distributed tracing on a single-service box is air-traffic control for a paper plane. What actually earns its place: centralized log search + RED metrics from the access log. No always-on OpenTelemetry, no Tempo. The complement to my Beszel setup.