Some notes on technical design documentation practices such as RFCs, ADRs, decision logs, and related approaches for software projects.
Discover gists
| POSTGRES_PASSWORD=replace-with-a-strong-password | |
| REPLICATION_PASSWORD=replace-with-a-different-strong-password | |
| ADMIN_PASSWORD=replace-with-another-strong-password |
| This guide explains how to install .NET Framework 4.0 and 4.5 on VS 2022, as it doesn't support these versions out of the box. | |
| Microsoft says these versions are outdated but many people still need them to maintain legacy apps. | |
| Requirements: | |
| - VS 2022 | |
| - 7-zip (I haven't tested WinRar) | |
| Notes: | |
| - When .NET 4.0 is mentioned, it means .NET Framework 4.0 | |
| - When it's mentioned to open as an archive or un-zip, use 7-zip |
| // SPDX-License-Identifier: MIT | |
| // | |
| // This program contains an implementation of SSIM-based perceptual image | |
| // downscaling for PNG images. | |
| // The program can be twice as fast when compiled with -Ofast. | |
| // The program behaviour can be adjusted with predefined preprocessor macros: | |
| // * -DTILEABLE: Assume images wrap around at corners. This should be enabled | |
| // when downscaling tileable textures. | |
| // * -DGAMMA_INCORRECT: Downscale without applying the sRGB EOTF and OETF. | |
| // When downscaling images with symbolic meaning, e.g. screenshots of text or |
Path inventory: 2026-01-21. Table inventories: 2026-06-10. macOS build was not recorded in the original inventory.
~ denotes the user's home directory. Paths and schemas vary by macOS version, account configuration, and library location. Listed columns are schema examples; their presence does not establish populated values or retention periods.
~/Library/Application Support/Knowledge/knowledgeC.db
A pattern for building personal knowledge bases using LLMs.
This is an idea file, it is designed to be copy pasted to your own LLM Agent (e.g. OpenAI Codex, Claude Code, OpenCode / Pi, or etc.). Its goal is to communicate the high level idea, but your agent will build out the specifics in collaboration with you.
Most people's experience with LLMs and documents looks like RAG: you upload a collection of files, the LLM retrieves relevant chunks at query time, and generates an answer. This works, but the LLM is rediscovering knowledge from scratch on every question. There's no accumulation. Ask a subtle question that requires synthesizing five documents, and the LLM has to find and piece together the relevant fragments every time. Nothing is built up. NotebookLM, ChatGPT file uploads, and most RAG systems work this way.
Apple is working on a Solar wallpaper engine for macOS 27 that groups aerial Landscape wallpapers and automatically switches variants based on the sun's position during the day.
When properly configured, the four Tahoe and Golden Gate landscape wallpapers appear under one entry each in Settings, and a new Automatic display option appears. Here's how to enable it.


