All the Posts

In June 2025 migrated this blog from Ghost to Hugo. Here’s why and how.

Our table legs were too short to allow me to use some donated drawers underneath for our nursery. Rather than using boring table leg extensions, I printed something fun instead. The feet were a great success, using AI was… less so.

Taking some time between jobs, I headed down to Lisbon to build my own wooden surfboard with Antoine at Ondanova surfboards.

Most of the focus of getting to 2.0.0 was on accuracy and robustness, however much of the focus since has been on performance and speed. This post is a little peek behind the curtain of how we went about that, and what some of the biggest improvements have been.

Today we released SQLFluff 1.0.0, the first stable release of SQLFluff.

A dive into PCIe and taking a dremel to a Razer eGPU enclosure in search of stable Gigabit Ethernet.

The information you have is not the information you want. The information you want is not the information you have.

…or at least can never be “purely” objective.

The kind of ‘courage’ I’m talking about at the end doesn’t need to be loud. Often courage is presented as just the knight on a horse leading the charge, but I think courage can also be a quiet confidence that doesn’t let an issue drop, or just asking the question that nobody is asking at the right time.

There is a balance to be played in the role of any person who is responsible for metrics and measurement. How to approach the perennial question: “can you make me a dashboard/report so I can see X”.

Once upon a time, there was a company of tailors… and so she drew up a position for a number wizard…

This year I spoke at the Coalesce 2021 conference about a topic a bit broader than just SQL linting or data in small businesses.

I printed a set of the Back-Loaded Horn Speaker V2.0 by guppyk from Thingiverse. They’re pretty good, but the I think I was pushing the capabilities of the printer a little far.

Stuck in another lockdown I embarked on building a guitar. Things came out pretty well.

In June, I had a great chat with the team at the Podcast.__init__ (aka the Python Podcast) about linting, SQL and open source projects.

We wanted a statement light for our new flat, but didn’t want to spend hundreds of pounds. I made one out of some scrap wood.

An old tabletop, some reclaimed construction timber and some very big dovetails.

The dbt project at tails.com has over 600 models and 66k lines of code. With multiple contributors to a project and varying SQL backgrounds, it’s really difficult to maintain consistent readability and comprehension across a codebase like that by hand.

I already feel less stressed.

PETG printing badly? Maybe you also need a dry box.

Following on from part 1, we go from getting serious into philosophy as we track our growing analytics function.

A rambling tour through life in small businesses and the importance of culture and leadership in data organisations.

Had some spare wood lying around and needed a new guitar stand. I try my hand at some traditional joinery.

Many people are rediscovering bread, and many organisations are rediscovering data. Both are things that people are devoting considerable time and attention to right now to provide reassurance, grounding and direction in what is otherwise a bewildering time.

For around a year I’ve been working on SQLFluff, a SQL linter to keep up with the growing place of analytics code as a store of business definitions. This post isn’t about why I wrote it, but it is about how to start making it a bit more appealing.

The Croydon wood recycling centre is a great place, given everyone is locked down during the Covid19 pandemic, it’s been a great time to make progress on some home improvement. This project involved… a lot of sanding.

I was very humbled to be included in the DataIQ 100 for 2020. This is a repost of the bio included in that.

My journey to upgrading Ghost and setting up continuous deployment for a custom Ghost theme using CircleCi.

SQL has been around for a long time, as a language for communicating with databases, as a communication protocol. More recently with the rise of data as a business function, or a domain in it’s own right SQL has also become an invaluable tool for defining the structure of data.

In this episode of the Hub and Spoken podcast, I have a chat with Jason Foster at Cynozure, discussing building data functions from scratch in a fast-paced business.

Alan presenting on the dbt setup at tails.com in one of the early London dbt meetups. Talking about continuous deployment, testing, alerting, and making charts on top of your dbt runs.

If you use dbt and Snowflake, we made a dbt package you can use to make sure that you keep your dependencies tidy and understandable.

If one screw doesn’t give you the stiffness you want, then why not add some more?

Designing a new, smaller, compact direct drive extruder for the hypercube was hard.

We set out back in August to build a CNC Mini Mill using an old tumble drier motor. Here’s how the plan changed.

Lots has been written about data functions with lots of tools and budget. In startup-land that’s not an option. This is the story of how we’ve done Business Analytics on a Shoestring.

How hard can it be. I mean, building a dasktop CNC Mini Mill from scratch, without access to an existing mill or lathe. If nothing else I’ll learn something along the way…

In Codeship, CI environment variables aren’t available at container build time. Here we use build arguments to achieve the same end and bake CI variables into containers.