Threading Revisited

Volume 8, Issue 70; 04 May 2005; last modified 08 Oct 2010

The overhead of maintaining threads was too high, so I've decided to do something else.

Maintaining thread information “by hand” turned out to be too difficult. Invariably, I'd write an essay and forget to update its threads, or I'd update some thread data but forget to rebuild the thread pages. (I never did implement a way to automatically build just the right threads.)

To fix these problems, I've rewritten the threading code to take a completely different approach: threads are now constructed automatically from the tags associated with an essay.

If Technorati wasn't still ignoring my tag data, I suppose I could just rely on their tag pages to provide the “threads”, but just because they ignore my tags mean I have to. And I can still include “off site” essays in the thread by dropping a little metadata into the system.

It's just as useful (or useless, your call) as before, it can still include essays written by others, and it's maintained automatically. Win, win, win. Oh, and the images work in the PDF versions now too. Win number four.

My latest thought is that I should probably unify threads and category feeds. That is, if there's a thread page, there should be a feed for those pages, and vice versa. But that's a bit of hackery for another day.

I really ought to get back to WITW.