Summer makeover

I've been meaning to go back to blogging for a while, now. One of the reasons that kept me from doing so was how cumbersome my blogging setup was; last November, I considered moving to Pagecord but remedy this, but then it was the cumbersomeness of migrating that held me back.

Well, not anymore.

Third time is the charm

This is the third iteration of my blog. It started as a Jekyll site, hosted on GitHub pages. The free hosting was the main factor in this decision; I only went with Jekyll because it was the path of least resistance. And I had a good opinion of Jekyll – but that opinion didn't last long. The templating language was very annoying to me, and Jekyll felt over-engineered to me.

Eventually, I switched over to Zenweb. In part because I love Ryan Davis' work, in part because Zenweb is the minimalist, Ruby-centric kind of tool that I like. It is also a bit rough around the edges, and getting it to generate exactly the kind of site that I wanted was a bit involving – but also a good experience. It got me to dive into interesting stuff. However, because I was still using GitHub Pages as my hosting solution, publishing a new post was always a chore. I had to generate the whole site, then copy it to a different repo, then push to GitHub… Too many steps. Fiction erodes motivation.

Last year, I stumbled upon Pagecord. Olly seems like a great guy and I felt like supporting him, and Pagecord is a really neat and polished piece of software. However, as it was, Pagecord was not completely a good fit for me. Besides the migration itself, moving over to Pagecord would have meant losing a few things that I cared about, and so I stayed – and let my blog rust away.

The importance of killing one's darlings

Over the past months, Pagecord gained features I needed to switch over, in particular the API. However, until yesterday, I was still blocked by a few things:

• I like to write in Markdown, in my text editor of choice by default, rather than in a rich text editor in a web browser, or an email client.
• I want to keep my local Markdown files as the single source of truth for my posts, rather than have them in someone else's database, even if they are easy to export.
• I like to use footnotes in my posts, and they are not supported by Pagecord's editor.
• I like to render code samples in a <figure> HTML tag, with a <figcaption> element to indicate the file the sampe is from, and this is not supported by Pagecord's editor either.

Fundamentally, moving over to Pagecord meant giving up some control over the way I write, and over the HTML code that exposes my writing on the Web. These were darlings I didn't want to kill.

And yet, I did. The main reason was that they were holding me back – between the hammer of an old pipeline so cumbersome that it discouraged me from writing, and the anvil of lost functionnalities. Another reason was that I really wanted to give Pagecord a proper chance. And, in the end, I decided that I had to kill my darlings because they were my darlings. Maybe I would enjoy writing in a web browser. Maybe I can do without footnotes. But I'll never know if I don't try.

How it went

Claude did a good job of importing all my content into Pagecord – even if, as always in my experience with coding agents, guiding it was as frustrating as it felt like magic. Pagecord being first and foremost a blog forced me to reconsider the organization of my content, and give up on the distrinctions between articles, posts, and other categories, which is for the best. (Another reason to kill these darlings.)

I still have to review the migrated posts by hand and fix the errors that can be fixed, like the broken links, or rendering issues with some code samples. I have accepted to give up on the footnotes and my custom styling, even if I was really proud of some of my CSS tricks.

I will keep the repo with my old content in Markdown form. In fact, I also used Claude to backport the organizational changes there. And I may export the new posts from Pagecord to the old repo ona regular basis, as a form of backup. We'll see.

The important thing is that, even if this new version of my blog is be less flexible and less bespoke than the previous one, it is also way simpler to use, and should get me back to blogging again.