The text discusses the process of un-double-encoding UTF-8, explaining the technical aspects and challenges involved in handling corrupted UTF-8 data. It provides insights into the methods used to correct these issues and the impo...
About:
Fluffy and Flakey is a blog that covers a variety of topics including cooking, engineering, programming, and more.
Website:
Outgoing Links:
Subscribe to RSS:
Detecting unclosed HTML attributes can be achieved by analyzing raw values through WordPress' HTML API, avoiding common pitfalls of false positives.
The author shares a detailed session on managing 'stacked branches' in git, demonstrating techniques for reorganizing commits, resolving merge conflicts, and maintaining a clean commit history. The session includes practical steps...
Learn how to efficiently track changes in specific lines of code using git's -L parameter, focusing on WordPress' package.json file.
The post discusses various methods to compare a git branch against its target branch, particularly when reviewing a pull request (PR). It highlights the challenges of using different git diff commands and introduces the --merge-ba...
5.1Observed software decay
2026-02-22 • technology technical issues outages
The post highlights a troubling trend of software and service failures, particularly with GitHub and macOS, suggesting a broader decay in internet reliability.
The text discusses the potential impact of large language models (LLMs) on various aspects of society and the software industry. It highlights concerns about the misuse of LLMs for surveillance, the potential for misinformation an...
0LLM Challenge: Infer context window (take 2)
2025-06-14 • censorship ai
The author discusses the difficulty of persuading an LLM to count to very large numbers and the conversation they had with DeepSeek-R1-0528 about it. The LLM refused to perform the task and instead provided fabricated responses.
The author discusses using llm, a tool for working with LLMs, and how they often forget to format the output visually. They created a fish function to wrap llm and make it easier to read.
The author is facing a challenge of inferring the maximum output token count for a given model without knowing the parameters in advance. They have tried various prompts but the models refuse to comply or truncate the response.
The post is a reflection on the author's participation in the CloudFest Hackathon 2025 with the CMSFreedom team, focusing on enhancing interoperability between content management systems. The team demonstrated a workable model for...
The text discusses the author's exploration of self-hosted LLM options, specifically llm-mlx, and the challenges faced in getting it to work locally. The author also provides an example of Erlang code to traverse a tree.
0Six common HTML myths.
2024-11-15 •
The text debunks six common myths about HTML, including the misconception that HTML parsers are loose and forgiving, that XHTML is preferable to HTML, and that balanced HTML tags can be parsed by regular expressions. It also expla...
0UGML: A proposal to update SGML.
2024-10-11 •
The author is proposing a new markup language called UGML, which is an updated version of SGML. UGML retains some characteristics of SGML while discarding outdated features. The author discusses the features of UGML, what it is no...
0My WordPress Contribution Workflow
2024-09-20 •
The post is about the author's typical setup for working with code, tools, GUIs, and helpers for git, and the process of committing to Core. The author shares their workflow and the tools they use, and how they prepare the commit ...
0Don’t convert HTML Text Encoding
2024-09-18 •
The text discusses the issues with converting HTML text encoding and the potential data loss that can occur. It explains that it's generally not safe to apply text-encoding conversion to an HTML document without preprocessing it, ...
0On SGML: a Summary
2024-09-02 •
The text discusses the SGML markup language, its impact on HTML and XML, and its features such as DTDs, syntax, and character references. It also reflects on the limitations of SGML and its potential for resolving markup syntax is...
0Today I was slopped in the face.
2026-01-15 •
...
0CSV files can be UTF-8.
2024-07-31 •
The text discusses the issue of CSV files encoded as UTF-8 displaying incorrectly in Excel on Windows and Mac. It explains the problem of double-encoding a string as UTF-8 and provides a solution using a Python script to add a Byt...
0Profiling WordPress, v6.6
2024-06-18 •
The author discusses the methodology for benchmarking the performance changes in the upcoming WordPress 6.6 release cycle. They explain the test setup, the process of finding the commits to test, creating an empty site for each co...
0Rooting out Gremlins
2024-03-21 •
The text discusses the phenomenon of gremlins in text, which are garbled junk characters that appear due to confusion in character encodings or special rules from the HTML specification. It explains the history of character encodi...
The text discusses the use of character references in source code and URLs in HTML. It explains how the same text may not work in different HTML documents and how the rendering may change the document's meaning. It also delves int...
0HTML: Scan all the tokens.
2023-12-22 •
The author discusses their mission to eradicate the use of regular expressions while processing HTML, and shares three tools they have built to help with their work. They also mention the internal refactor that opens up scanning a...