Giles' blog

Going through the archives

I've been looking through my archives over the last few days while laid up with a minor medical problem. Along the way, I stumbled across a draft from 2021 -- written just after I moved my blog from Wordpress to a static site generator -- that never saw the light of day.

I really don't know why I didn't finish and publish it at the time -- it's got some good stuff. And in the light of my decision to write more TIL deep dive posts, it's particularly relevant now. So here it is, dusted off and lightly updated.

One of the main changes in moving this blog over to my new static site generator has been the change in the format for the posts. I wrote it so that it could still render (reasonably well) the semi-HTML that Wordpress uses to save posts, but its main source language is Markdown, with which I can use code fences for syntax-highlighted code blocks, and a number of other nice typographical tricks -- and posts look much better with a few changes to take advantage of that.

So over the last few days, I’ve spent more hours than I probably should have going through all of my old posts and hand-converting them to Markdown. There was absolutely no need to do this, but it felt like the right thing to do. A touch of OCD? Well, possibly, but there have been other benefits.

I was able to read through the history of the rise and fall of Resolver Systems, and the resulting rise of PythonAnywhere; it was very noticeable how the number of posts per month rose as things became more precarious with Resolver, reached a fever pitch just as it was becoming clear that we just weren't going to make it, then practically fell off a cliff when we pivoted to PythonAnywhere and quite rapidly realised that it was actually going to work. It turns out that having a popular and growing product leaves less time to blog, surprisingly enough.

(I was also surprised at how cool our Dirigible cloud spreadsheet was -- it's a pity that we never had the time to explore it further, though I think that given the resources we had at the time, it was right to focus on PythonAnywhere.)

I was also able to see what I've posted in the past and get some kind of feel for what seemed to work, what didn't, and what was just terrible and pointless.

Good kinds of posts -- I should do more of these:

• Project writeups with detailed notes about what I did. These were really fun to write and are really fun to read now (at least to me), and tend to get lots of comments, which is a nice dopamine hit but more importantly means that people are reading them and getting value out of them. Even short(ish) work-related ones are worth re-reading.
• Posts about weird problems I've encountered with fixes. These are, I think, useful if well-written enough that someone searching for a solution will find it.
• Funny quotes, so long as the excerpt contains enough to be amusing on its own. The original link might die, but with some context they can still be funny.
• Thoughtful posts about actually interesting things happening with my day job; most of my posts about Resolver Systems are dull to read now and were probably dull to read then -- tech company founder boosterism -- but at least one was written in my own voice expressing exactly what I thought, and it's really fun to read. I particularly like the "Client being acquired by a non-bankrupt competitor, all expenditure on hold. End user still keen" quote.

Indifferent kinds of posts:

• Quick links to interesting things which will probably disappear from the Internet within a year or so. I'm torn on these. On the one hand, it's nice to keep a list of links that's not at the mercy of a social media platform like Twitter/X. On the other hand, dead links are a waste of space, and reading through them is kind of depressing.
• My Terribly Important Thoughts on Something Unrelated to my Expertise. Is there any point in posting these? They feel a bit self-indulgent, like they're posts I'm writing for my own benefit and not for the reader's. But maybe they're of some interest to some reader somewhere?

Bad kinds of posts:

• Thinly-disguised advertising for my day job. This blog has 34 posts in the Resolver One category in four years. Sheesh. Compare that to 16 for PythonAnywhere in a decade. A good product doesn't need its founder hyping it up non-stop!

So that's what I've found. Will knowing that all change how I blog in the future? No promises -- but at least I have a roadmap now.