Mastodawn

Most web projects are still pages, forms, users, records, uploads, emails, reports, and scheduled jobs. For that shape of work, LAMP is still a practical choice. Old Stack Journal: https://oldstackjournal.com/web-tech/lamp-is-still-enough-for-most-web-projects/?utm_source=mastodon

Not every side project needs a modern JavaScript stack.

For a lot of small tools, PHP + MySQL + Apache + cron is still enough. If the job is forms, tables, and scheduled updates, keep the stack small.

If you keep typing the same setup prompt into ChatGPT, a custom GPT can be a practical way to save time. OSJ’s take: less repetition, less drift, more reusable workflow. https://oldstackjournal.com/workflows/why-you-should-build-your-own-gpt-instead-of-repeating-yourself-forever/?utm_source=mastodon

Pointing a domain at a home server is not one job. It’s DNS, port forwarding, Apache, and sometimes CGNAT or a dynamic IP getting in the way. This guide breaks the setup into parts and keeps the public surface area small. https://oldstackjournal.com/how-tos-guides/point-a-domain-at-your-home-server/?utm_source=mastodon

Practical n8n notes from a real workflow: HTTP Request, Wait node, webhook callback, human review, continue. I also wrote about where Automation Receipts fits beside it as the receipt/review layer. https://oldstackjournal.com/?p=241&utm_source=mastodon&utm_source=mastodon
#n8n #Automation

EVE Online haulers and market runners: we built Eve Trade Routes at https://terralog.online/.

Decision-first route checking before you undock, with a practical focus on route planning.

Old PC or mini PC? You can turn it into a useful Ubuntu home server for local web dev. This guide covers Ubuntu Server, Apache, MySQL, PHP, SSL, local domains, and safer outside access. Practical home lab stuff, not enterprise noise. https://oldstackjournal.com/how-tos-guides/how-to-set-up-a-fresh-ubuntu-home-server-for-local-web-development/?utm_source=mastodon

Is it down...yes, yes it is.

Vibe coding is useful, but the builder still has to steer. Current files, small changes, full replacement files, test after each patch. That’s the workflow that actually holds up. OSJ note: https://oldstackjournal.com/vibe-coding/vibe-coding-is-useful-but-you-still-have-to-be-the-builder/?utm_source=mastodon

Most bad LLM prompts fail because they leave too much open.

I wrote a practical guide on prompting ChatGPT, Claude, and other LLMs for coding, debugging, writing, and small web projects.

Give the model the job, context, rules, examples, and a clear definition of done.