Order of the Bath

In the last few days I have been hit by comment spam—rotten really as I don’t ordinarily get many comments at all. You won’t have seen them because there are two lines of defence. You first have to register an account and even then the comment is sent to me first for moderation. This probably puts legitimate comentators off but I see I was right in chosing this system.

The first of these bulwarks is not working. Somehow the spammers are getting comments into the system without first registering accounts. It seems to be a timely reminder that I need to upgrade to WordPress 2 so please excuse me if the blog is down for some of the time over the weekend.

A secondary benefit is that I am told that Akismet spam defence is built into the new version, so perhaps I may be able to relax the rules a bit, but one thing at a time.

An RSS feed is an XML document with an implicit DTD. Unlike XHTML, this DTD doesn’t specify any named character entities so the only ones allowed are &, ', ", <, and >. This means that feed generators, like the one in WordPress and other blog & content management systems should convert any other character entities into numeric form before serving the feed document. WordPress 1.5, at least, doesn’t.

As a result, the feed received by the reader may generate a syntax error. In practice, most don’t seem to notice entities in the description text and pass them straight through for rendering but some, including Feedreader, choke if there are any in the title.

Conclusion—don’t use named entities in your blog titles.

Well, that was hard work; but I’m quite pleased with the result. The apparent simplicity hides some rather complex stuff going on underneath. I can’t imagine any non-geek being able to customise their blog but perhaps they use a simpler but less flexible system. Even I’m glad that PHP is a pretty obvious programming language—not a huge step from sh or perl. The tricky bit is the CSS—it is still very much a black art to me.

I have just commented out the unwanted stuff from the default template for the moment but once I am happy with it I will cut out the redundant stuff which should speed it up a bit.