The blog continues to evolve in structure and design to make it easier to generate the blog (and maybe, finally, enable rss).
While trying to figure out what I need to generate RSS, I realized that I had a bunch of publish-date data inside the files (i.e., “Published: YYYY-MM-DD”) that I would have to parse and extract to emit the files in correct order. Similarly, I had no way to generate the index page when I add new posts. I couldn’t use filesystem access or creation metadata because these files move around a lot. I also couldn’t use git data for the same reasons.
The fix was to move all posts into a Y/M/D path organization and treat the path as the sorting key for the index. Full details of my changes start at cdf2896 and end around 671a4d7. If I post more than once per day, I can add filenames like “1_foo”, “2_bar”, and so on to retain proper sorting.
Update: I, in fact, did this today. It works!
Now that the files are organized by publish date, I’m investigating what it takes to reprocess the posts into a full-text rss feed.
More analysis and data to come.