25/08/2002-26/08/2002 @01:22:55 ^01:49:19

In an effort to code something that isn't a gratuitous security hole (please don't click on that link if you're at work! Or even if not! Trust me!) we made this, a new file format:

        <!-- BLOG entrytime uploadtime -->
valid standalone HTML fragment

This is repeated as necessary. "Standalone" means that it will go anywhere in an HTML document and not break it. So you could have a whole paragraph, or table or whatever. But a fragment of a table isn't standalone, somewhere else would have to insert the <table></table> tag pair.

Although this file format is not actually valid HTML (there's no DTD, header, body, title...) most browsers will still attempt to render it, so you could write it and still be able to check your links are correct and how everything lines up and so forth.

Then you have some scripts that get each fragment and plonk them into a template to make a valid document. The dates, which are EPOCH TIMES i.e. easily computer-readable, are present for the scripts to make headers for each entry and also for sorting into the correct file when the time comes to archive old material. This probably won't be dynamic but controlled by a Make file when the site content is written. It's more likely not to be a gross disgusting security hole that way (you've had enough disclaimers by now)

That's all very well but so far all I have is code to kludge the entries I've already got in this file and the archives out of them and into this new format. It's not pretty. There are all sorts of things that it needs to handle, spurious use of local time where GMT time would have been better, nonconsistent presence of upload times, the format of the <h1></h1> line changing randomly around the start of April, you name it...