Wherein we speak of tags, categories, and content. We also touch briefly on the general inadequacy of GitHub’s implementation of Jekyll, and at greater length on the need for a build system.

Progress:

The site is actually beginning to come together. The “news” category has been created, so that the home page can show only a limited set of blog entries. There is a generic include file for blog entry metadata – in Liquid, include files take the place of functions – you can pass parameters to them. But if you want to pass the current page (for example) you have to assign it to a variable first. Liquid isn’t the ugliest programming language I’ve seen (you don’t want to know–really), but it leans in that direction.

I’ve also added post navigation in the form of “next” and “previous” links. Jekyll really is fairly blog-aware. With one major exception: tag pages.

Regress:

One would really like a page for each tag used in one’s blog, that lists all the posts with that tag. The problem is that Liquid has no way to construct pages on the fly. There are apparently some plugins that can do that, but GitHub doesn’t support them. People have developed three solutions:

  • The pure Liquid solution (pun acknowledged) is to generate a page that simply lists all posts, grouped by tag. Not too bad, actually, and with this site’s current meager set of posts, it would work okay. It’s definitely the simplest. There is a version of it here on StackOverflow
  • Manually make a page for each new tag. It doesn’t have to have any actual content, just front matter with the layout and the tag name.
  • Grovel (that’s a technical term – it’s in the Jargon File) over the site with a Python or Ruby script, find all the tags, and build the pages on the development machine.

There are a couple of other reasons why one might want a proper build system in the development environment – it would be useful, for example, for creating posts. It’s not as if I haven’t done that before!

Another reason to have either a build system or some utility scripts would be to suck over content from other places; e.g. the old site, various other places where I have relevant documents stashed, and my Dreamwidth post archive. (That currently has nearly 13,000 posts in it. I suspect that Jekyll would have a hard time handling it, since every tag page will involve iterating over the entire set to filter by tag.)