Books and websites and clouds, oh my!
And totally failing to stick to "one thing at a time"
[The amount of time between newsletters continues to amuse and embarrass me in equal measures. I knew I had a draft newsletter that I wanted to finish and send out before the end of the year. I was surprised to find the draft dated from the middle of September!]
The title of this newsletter is supposed to remind me that it’s easier to make progress on a project if you only work on one thing at a time. The contents of this edition will be an object lesson in how bad I am at sticking to that.
Data Munging with Perl (2ed)
I am still making progress on this. But I’m embarrassed when I think of the talk I gave about it at the last London Perl Workshop, where I suggested it might be available at around Christmas. Luckily, I didn’t say which Christmas.
I have made progress on it though. In particular:
Working on the chapter that used to be called “Hierarchical Data”. It used to cover HTML and XML. The new version will cut down on the XML content and replace it with coverage of JSON and YAML. I’ll publish a blog post soon that contains an extract from the new version of this chapter.
I got a few reports from LeanPub that the “work in progress” version isn’t working very well for some people. I dug into that and it turns out that the EPUB document that I saved from Google Docs isn’t a particularly standards-compliant EPUB. I’ve been working on a new production pipeline where I download a DOCX version and convert that to EPUB using Pandoc. That seems to be working better.
And let me briefly remind you of the existence of the work-in-progress version. You can buy that now from LeanPub and you’ll get access to all updates - including the finished version when it arrives. That’s very useful for me, as I get to iron out little problems like the one I described above before I publish the final version.
Design Patterns in Modern Perl
I did publish a book, though. My good friend, Mohammad Sajid Anwar, approached me in the autumn and asked if I would be interested in publishing his book, Design Patterns in Modern Perl. If you know Mohammad’s work from Perl Weekly or The Weekly Challenge, then you’ll understand why I leapt at the chance.
We managed to get the book published on both LeanPub and Amazon just in time for Mohammad to announce it at the end of his talk at the London Perl Workshop. In working on turning his Markdown into an ebook, I took the opportunity to update my rather dated publication pipeline, and I wrote a blog post about that - Behind the Scenes at Perl School Publishing.
Mohammad tells me his next book is almost ready (and he has another in progress behind that). I suspect we’ll be renaming Perl School to “Anwar Books” before the end of 2026.
Improving websites
If you’ve followed my work for a while, then you’ll know that I run a lot of websites. At some point, they were each going to be the big, successful project that would make me ridiculous amounts of money and stop me from needing to work for a living. Of course, it never quite worked out like that.
Part of the problem (or, at least, as far as I can see) is my limitations as a web designer. I was delighted ten years ago when I discovered Bootstrap and realised my websites didn’t need to look quite as bad as they originally did. But, eventually, I began to realise that while my sites no longer made me want to scratch my eyes out, their design still left a lot to be desired.
Luckily, ChatGPT is pretty good at website design. So, together, we started a project to improve the look of many of my websites. Let me show you an example.
Read A Booker
This is how Read A Booker looked a few months ago. It’s a site that lists all of the shortlisted titles for the Booker Prize over the years. Of course, it’s just a way to encourage people to buy the books using my Amazon Associates tag. But it doesn’t really work when the site looks as uninteresting as that.
And here’s what it looks like today. Nothing has really changed in the structure of the site, but it just looks that little more enticing. It looks like a “real” website (whatever that means). The other pages all have similar improvements.
Oh, and as part of that project, ChatGPT also helped me write a little Javascript library that adjusts Amazon links and buttons so they automatically go to a visitor’s nearest Amazon site.
Line of Succession
ChatGPT has also helped me make some improvements to the Line of Succession website. This is a site that started as an intellectual puzzle and has grown into my most successful website. It’s a site that allows you to pick a date in the last 200 years and will show you the line of succession to the British throne on that date. It’s pretty niche, but it gets a reasonable amount of traffic each month. The British royal family are obviously still of great interest to a large number of people.
Over the last few months, my work on this site has been in three main areas:
The site has been given a lick of paint. It’s still Bootstrap underneath, but it now looks far more professional
I’ve done some work on the underlying data. Obviously, I know from the news when someone near the top of the list is born or dies. But it would be easy to miss changes further down the line. With help from ChatGPT, I’ve developed some code that queries WikiData and looks for births, deaths and other changes that aren’t already in my database. Later on, I think I can reuse a lot of that code to push my data further back in time
The site is driven by a Dancer2 app that queries an SQLite database and builds the line of succession for a given date. This isn’t particularly efficient and I’ve started redesigning it so the line of succession is precomputed and stored in the database. This will make the site much quicker. This work is ongoing
Moving to the cloud
I still run a lot of my web apps on a VPS. This makes me feel like a dinosaur. So I’ve been moving some of those apps into the cloud. I wrote a blog post about my experiences.
And then, because some people aren’t starting with nice, PSGI-compliant apps, I wrote another blog post explaining how you could also do that with crufty old CGI programs.
Some silly toys
I’ve written before about how AI-assisted programming means a lot of little projects I wouldn’t previously have had time for are getting ticked off. Another version of that is fast becoming a good way to get bugs fixed in my software:
Raise an issue in the GitHub repo
Assign the issue to Co-Pilot
Wait 20-30 minutes for the pull request to arrive
And most of the time, I’m finding that those PRs are of very high quality. Any problems can often be traced to deficiencies in my specification :-)
Here are a couple of toys that I’ve recently written that someone else might find useful:
Dave’s Dry January Tracker - an older project, but it’s had a revamp for 2026
MyDomains - I’m getting over my domain buying habit, honestly! But this helps me track the ones I still have
Feeding my soul
But I can’t work all the time. I need to soak in a bit of culture too - whether that’s consuming or creating. And I have plans to do more of both next year.
Consuming
Film - for some reason I’ve seen about half as many films in 2025 as I did in 2024. I just got out of the habit. It’s a habit I plan to resurrect in 2026. Feel free to follow me on Letterboxd.
Books - I used to be a voracious reader, but I’ve barely finished a book since the pandemic in 2020. You can follow my attempts to read more on Goodreads.
Creating
I’ve written a couple of short stories that I’ve published on Medium.
I’m experimenting with a long-form piece of fiction based on The War of the Worlds.
I’m no musician. But I’m enjoying the possibilities that AI music creation is bringing me. My main project is an artist called Oneirina (that’s her at the top of this newsletter). She’s on Spotify, too. I don’t pretend it’s deep or meaningful, but I think it’s fun.
Anyway, that’s a wrap on 2025 (especially for those of you who are in Australia or New Zealand), I think. I’ll see you in 2026.
Happy New Year,
Dave…