Some of you may have been turned away from pages on the site by ugly validation error messages that killed the whole document. For once it's those with the decent browsers who are more likely to have had problems.

It's because the W3C's validator pointed out to me that text/html is not a registered MIME type for an XHTML 1.1 document and that it should be application/xhtml+xml. I changed the content type of my 'blog pages.

The reason application/xhtml+xml is considered better than text/html is that it properly represents the XHTML 1.1 as an XML application, and user agents should validate it as such. That's great, except that a browser with decent XHTML support given a document of type application/xhtml+xml should stop parsing on a validation error, hence the big ugly messages. It's fine while my entries are valid, but the moment I put a foot wrong - bad quotes, unescaped ampersands in href query-strings, etc. - the whole page becomes unviewable.

I've reverted the site to XHTML 1.0 Strict. It's much the same set of rules, but text/html is a valid content type again. It was nice to ride the bleeding edge for a while, but if a standard leaves me no room for error then it's clearly not the right one for my personal 'blog.