At first, I thought everything at twoprops.net was lost. Until it happened, I hadn’t appreciated just how invested I was in this project. When I realized that the backups hadn’t been working, I was devastated.
I had been editing the home page (index) at the time, so my editor at least had a copy of that, listing all the columns I had ever published by their title, date, and description.
Eventually, I discovered that the backups had been working until I moved the server to the island. That meant that everything from March to late August could be restored. It was a bit of a process, but the restoration was complete. Unfortunately, that still left almost 100 lost columns.
I checked the Wayback Machine at the Internet Archive and discovered they had captured only about 17 of those lost columns. Pretty disappointing.
I then tried a hacker’s trick of searching disk space on my various machines for keywords to see if anything from twoprops.net got cached, and hit the jackpot. I use Thunderbird as my main desktop email client. When I was researching RSS readers (note that that column has not yet been restored, but soon), I discovered that Thunderbird had an RSS reader built-in. I enabled it just so I could check how twoprops.net columns look in it. I’d been deleting each column as I read it, but it turns out that Thunderbird had cached all those deleted columns. Of course, I didn’t enable Thunderbird until mid-October, so I assumed all the columns from late August to mid-October were lost.
To my surprise, when I started trying to restore columns chronologically starting from late August, I discovered that they all seemed to be in Thunderbird’s cache. Then I remembered that I have twoprops.net configured to include the last fifty columns in the RSS feed… so when I first subscribed Thunderbird to my feed, it grabbed the previous 50 columns. That’s everything I’ve published since the backups stopped working.
The restoration takes time, though. I have to go through the twoprops process:
- grab the metadata for the column from the home page
- open a template file
- type in the metadata
- title
- description
- tags (these won’t exactly match what I had before but hey…)
- publication date
- previous column link
- next column link
- dump the text from the cached column file in the terminal
- copy and paste the html from the cached column into the template
- repeat for the next column
Fortunately, quartz is happy to just work with the cached html data in the vast majority of cases, so I won’t have to convert it back to Markdown. It will just be a bit more hassle if I ever have to edit any of those columns.
So I should be able — slowly — to recreate the whole site. Some tags will change, and some edits and addenda will be lost. I don’t think anybody but me will ever know or care.
—2p