In putting the finishing touches on an RSS 2.0 feed for, I've come up against a couple of issues surrounding the contents of the individual posts' <description> elements.

It seems to be pretty common practice to include entity-encoded HTML in the description. Not everyone likes this. So far, in trying out various desktop and web-based readers, I've enjoyed the HTML, although Mark Pilgrim's recent highlighting of security issues has me thinking a bit. On the other hand, I haven't tried reading RSS on my Pocket PC yet.

The other issue is full versus partial content for long posts. Should the descriptions be one paragraph teasers that require you to visit the site (like ongoing) to read the rest of the post, or should the whole post be in there (like Scripting News). As a consumer of RSS, I feel as though I would like some control of that; for some sites, like ongoing, I always want to read the whole post, but for others, like Wired News, I might only want to scan the summaries.

It seems to me that I should be producing four feeds — all combinations of both settings, and letting the consumer choose. Or maybe there's some RSS feature that I'm overlooking.