Four Things That Happened

One

Ina is one of the developers on my team, and a while ago she set up Lunch-and-Learn sessions on Friday afternoons. This is a session where we watch a recorded conference talk and have a discussion afterwards. The programme has a good balance of technological talks (on topics like infrastructure) and talks about the equally important people-oriented side of our industry.

Last Friday’s talk was Tanya Reilly talking about being glue - it’s a pretty heavy and hard-hitting talk about the dichotomy of what’s seen as “promotable” work and what work is actually necessary for a team to perform well. The latter is “glue work”, such as making sure the right people talk to eachother, making sure documentation is done properly, frequently talking to customers.

While absolutely essential glue work is rarely seen as promotable, and this has implications for the diversity balance within software development.

I won’t give more of a summary than that, and I highly recommend you watch the talk.

Two

I’ve finally taken the jump to move away from Google mail for personal stuff and signed up for ProtonMail. The lowest non-free tier is 4€ / month which is well within a sensible budget, and I can finally get my email behind <user> at probablyfine dot co dot uk

The transition was close to painless, the longest part of the process was me failing to set up some TXT records in AWS Route 53 properly.

I’ve set aside a bit of time for myself to turn these DNS records into an open-source Terraform module in the near future.

Three

I attempted to extract some interesting dimensions from some of work’s production git repositories during my 20% time, one of our shared code bases and a handful of larger, team-specific code bases.

1. Log commit_hash,commit_date,commit_message and additions/deletions for each commit.
2. Process commit_message using conventional commit guidelines into three new columns (type, scope, raw)
3. Join (1) and (2) into a nice big CSV

I wrote a small Java app for (1) and (2) which munges the output of git log, parses out the commit message using my conventional-commit library, and dumps it all out as a CSV.

Using a publicly-visible commit as an example, it turns

6afa2c4,2019-01-18,chore(base-nrpe): Improve lint-and-test.sh output.
6       3       lint-and-test.sh

into a CSV containing:

Running this app over 6 code bases, on commits from 2017-01-01 until today gives me a dataset with around 55k entries. The dataset is limited to 2017 onwards because that was when we first started using conventional commits at Unruly and that gives us two more dimensions to break down by.

I’ve already done a few quick calculations which I’ll be writing about in the near future.

Four

Stephen, another developer on my team, has been working on building Slack bots in his 20% time. He’s spiked and demoed a bot that bridges our Slack instance and AWS users, using a DynamoDB backend for lookups.

This is particularly interesting because there’s not a 1:1 mapping between username in AWS (enforced by Shift) and username in Slack (chosen by the user), so he’s built a layer of indirection to translate between the two when posting to the Slack API.

This is quite exciting because we can dispatch messages to both teams and individual users about the status of systems running in AWS.

Other stuff that happened

I finally finished reading Stephen King’s The Shining - I’ve not seen the film version but I know enough of the plot points that I didn’t find the book too scary. In the future I’ll be looking for books to give me a good spook, so if you have any recommendations, tweet them at me!

Baking is also back on the cards, as a way of relaxing at the weekends. I made a good old Victoria Sponge and scones in the last week. This was the first time in several years that I used an electric hand whisk rather than beating the cake batter by hand - and I’m never going back.