unbeldi, which I’ve spoken about before, is a very, very nasty hack that pulls data from a variety of sources and aggregates them to generate content into my website. I consistently bemoan it because it’s a pain and really shouldn’t work at all.

Today I’m working on a major update to unbeldi: removing the requirement to generate only one file. Currently, unbeldi checks for the existence of a file …/_posts/unbeldi/2013-10-05-index.markdown; if it exists, it won’t touch it (and thus won’t update it, if the input sources have changed), and if it doesn’t exist, it generates it. This is crippling, and makes testing and prototyping agonising.

How to do this is currently what I’m trying to work out. Ideally, I want to push the page stitching out into Jekyll, and make unbeldi just deal with the content aggregation. Eventually, this will make it easier to annotate content, especially content I aggregate in from Feedly, and it’ll make it easier to link to a particular article.

Jekyll supports pagination, but it’s not flexible enough for what I want to do: it supports a number of posts per page, which is unhelpful. Instead, I’m going to doctor up a solution similar to the one described here on StackOverflow. Preliminary testing suggests what I’ve done works quite well, too, so today should be the first day where this happens properly.

Current pages will all have to be sliced apart, annoyingly, and I’ll need to work out if I keep my current slug naming scheme (which astute readers will have detected as being approximately $SOURCE-%H%M%S. Hopefully I’ll whip up a script to make that work.

While I’ve been reading around, I’ve worked out, I think, how to remove the tag management crapheap, but I’ll leave that for after I fix unbeldi.