Other articles

1. My Chinese birthdays, time keeping is hard!

   I grew up in both the UK and China. So I'd like to think I have a little understanding of both cultures. China uses both the western gregorian calendar and a version of lunar calendar called 农历. This means I have a Chinese lunar birthday as well as a …

   read more

2. Asyncio backpressure - Processing lots of tasks in parallel

   There's mixed feedback to Asyncio in the community. Some people passionately hate it whilst others believe "writing async make program go fast". This debate is way too much for me to cover here right now. Though maybe I'll look at it in the future.

   For me I use asyncio a …

   read more

3. Docker's MCP toolkit First Look

   Short update since First Look at MCP. Docker has released it's MCP toolkit which includes a catalog of MCP servers as well as helpers to connect to the MCP clients.

   It is exceeding easy to use it, docker is registered as a single MCP server for your client. And tools …

   read more

4. First Look at MCP

   Previously I've played with tool calling in Langchain and Python sandboxes. But recently MCP (Model Context Protocol) is front and center. So I tried to create my own github MCP in Python to integrate cursor with the github cli.

   What is MCP?

   Before we get started, its good to familiarise …

   read more

5. Python 3.14: State of free threading

   In my posts earlier this year I talked about the parallelism performance on 3.13 free-threaded builds. In particular I looked at solving an advent of code problem. In How free are threads in Python now? I discovered significant performance penalties for using free-threading and a lack of tooling available …

   read more

6. t-strings: the good and the ugly

   This one's hot off the press as the first beta for Python 3.14 (aka. π-thon) has hit. We're looking at a chunky release with a lot of new features. But all I can think about are these new template strings (officially t-strings).

   PEP-750 officially introduces the concept. The idea …

   read more

7. Why you should write your tools in Python Again

   You're probably thinking that there are already plenty of tools written in Python.

   But I see that most of the popular tools like mypy and flake8 are built for development environments. In contrast, general purpose cli tools tend to be built in other languages, for example most of the docker …

   read more

8. Building a DSL with Python Operators

   I've been a little obsessed with operator overloading lately. First using |= in sqlalchemy-builder and then using | and @ in better-functools.

   I didn't actually know that these qualify as DSLs, specifically what's known as "internal DSLs". Funnily enough, I'm usually not a fan of DSLs. A few reasons come to mind:

   • DSLs …

   read more

9. better-functools: Python functional fun

   I recently put some effort into creating better-functools. It's a package that adds some tooling for functional programming in Python. And allows us to write some unique looking code in a manner similar to functional languages:

   Try in sandbox

   Why?

   At a glance this doesn't look very "Pythonic". More complex …

   read more