rebuilding the site

This blog started in the year 2000. I’ve gone through periods where I’d post something short every other day, and other periods where I’d post only one long essay every six months. But until recently the last post was in 2024. And then I was doing some work on the site in early 2025 when I broke it bad.

I was able to recover things to the point that at least my old posts would appear, but I lost the ability to log in and add anything new, or upgrade the WordPress software that ran the site, or do any maintenance on the site.

The great transformation

I had thought for a while of converting my WordPress site to Jekyll. Jekyll is an open source tool for making static sites. “Static site” just means that all the pages are pre-generated, so when you click to view one it comes up instantly, or at least as fast as your internet connection will allow. This is a different approach than WordPress, where when you click on view a page the web server software does a bunch of computations and generates the page in real time before showing it to you.

For a long time - maybe 20 years? - the site had used WordPress. When it came out, WordPress was a cool idea. It made writing new content and maintaining a site really simple for non-technical people. But it also requires paying for a web host service that can host WordPress sites. And that costs anywhere from $80 to $300 a year. Most of the time, I didn’t feel like I was getting my money’s worth. Plus, I’ve been increasingly dissatisfied with my web host. So I’ve wanted to switch to Jekyll for a long time.

I spent most of 2025 working toward that. It was a huge effort, because the remains of my site were in such a shambles. I couldn’t download the raw content - the text of the blog posts. I could only save the HTML pages that WordPress generated. And those had been generated by several generations of the WordPress software, so there was a lot of inconsistency.

But hey, I’m a software engineer. So I wrote some software that would take the downloaded HTML pages, pull out just the parts I wanted from each page, and save it in Markdown format, which is what Jekyll uses. It sorta worked, but had a lot of problems. And then “AI” came along. Once I started to play with Google Gemini CLI, I realized it could fix these problems way faster than I could.

So at that point, progress accelerated. I have racked up hundreds of dollars of “AI” fees from Google, but it’s saved me probably 120 hours of my own time. Anyhow, over the Christmas/New Years period, I finally recovered enough of my site that I could start to see it working in Jekyll. At that point, I switched gears from having Gemini write code to transform the old pages into the new pages, to having Gemini write code to clean up the remaining problems. That’s much easier. We’re not really done, but you can see the results right here. “You’re soaking in it!” as the old Palmolive TV commercial used to say.

What now?

I have a laundry list of problems with the site. But at this point I can start to add new content - new blog posts - and work on fixing old content at the same time. One of the biggest problems is that hundreds of images got lost when my site crashed in early 2025 and I haven’t been able to locate them anywhere. So you’ll find a lot of bad links on the site, and some pages that are photo-heavy are just a bombed-out husk.