Where Everybody's Crazy

Looks like I found a perfect approach to combatting comment spam...

I'm expanding Glob, my blogging software, to enable me to blog in two languages. I really respect Martin and Hazel and their bilingual blog, and I want to do the same. Not just for the integrity of knowing that you're not "talking about anyone behind their back", but also because I have a bunch of Japanese friends who get my blog (syndicated via Mixi, a Japanese social networking site) and would probably find it easier to read in Japanese than in English. Also it'll be good for my language skills!

I've done the first stage of hacking, which makes it a bit easier for me to create and edit blog posts, but now I have to think how to do the data model: not all my posts will be in English and Japanese - some will be just English, some just Japanese, some both. There should be a preferred language, and the option to switch. Don't worry, it'll default to English if Japanese isn't selected. :) Expect falling masonry for the next day or two...

    mysql> delete from tag;
    Query OK, 346 rows affected (0.00 sec)
    mysql> delete from tagging;
    Query OK, 2905 rows affected (0.00 sec)
    mysql> delete from thing;
    ERROR 1146: Table 'memories.thing' doesn't exist

"Wait, memories.thing? Oops. Wrong database"

OK, so I was playing around with a new toy and I trashed all the Memories tags. I've recovered most of mine, but now I have to work on recovering all the others.

The new toy was something I was working on just for a bit of fun. I just wanted to spend the afternoon coding, really. So I have started writing soemthing which aggregates tags from all of my tag databases: at the moment, I'm using tags for delicious, Memories and this very blog. It knows about all these. Hopefully I shall also teach it about other things which have tags, when I think of them. So, at last, you can see both what I was blogging and what photos I was taking on 19th March 2004 at the same time. Similarly, all my thoughts on Japan and on mission on the same page.

It's pretty raw and ready at the moment, but the eventual aim is to work backwards and have Memories, Glob and some URL-remembering-thing all writing their tag information to a common source, and, really eventually, querying the common tag database instead of having all the "find all Xs with tag Y and Z" code reimplemented for everything which uses tagging.

Incidentally, though it looks simple, the guts of it are beautiful. Each "thing" has a class, which is a privileged tag telling it where to find methods to, for instance, display itself as HTML. Tags have two components, a value and a style, allowing a use mention distinction. So a class tag class:memories would have a value of memories and a style - which is itself a tag - of meta:class.

meta:class's style is, of course, the recursive tag meta:meta. Did I just say "recursive tag"? Urgh.

I've been reinventing a lot of wheels recently, and I finally got around to reinventing the blog software wheel once again. There's good reason for this: I didn't like the way that Bryar handled tags (it didn't); I wanted to move away from flat-files into SQL so that I could edit things more easily, particularly sweeping up spam comments, which Bryar was terrible at; and I wanted to continue being able to syndicate XML feeds as on the right hand side. And, as usual, I don't trust other people's software.

The new software, Glob, is, well, a couple of hours old. I still need to tweak a few things about it - tag searching is still not there, for instance - and there may be some brokenness as a result of the changes, but this is all part of the fun. When it's all done, you'll have the ability to subscribe to particular tag streams - so you can only read technical posts, or only theological ones, or whatever. But for now all the old URLs are (should be) supported.

I took the opportunity to give the site a bit of a redesign as well.

This is a test of the glob posting program.