This is a simple, easy to install, self contained script for generating XML data.
Generally speaking, I almost always find it much easier to just use a text editor. For example, the ideal "content manager" in my case is the file system itself. (perhaps with CVS or other mechanism to provide automatic transfers and backups and a bit of PHP to ease the task of headers and footers.)
The upshot of this philosophy is that you can apply your tools and experiences to other areas.
Static RSS feeds would be ideal, except that by their very nature, they
need to be updated and the URL's need to be absolute. You can't for
example specify ../somefile.html as a URL, because the
reader doesn't know what it's relative to.
The lack of relative URL's is a real problem when working on a local development host, as the host names (and often path names) change.
The redundancy of data can also be a nuisance, for example the <link> tag and the <guid> tags probably need the same bit of data.
Being someone who almost always prefers a text editor to any GUI based tool, someone who hates content managers, control panels and other gimmicks and someone who likes to make use of standard unix tools, I wrote an RSS cooker that enables me to use a straight text editor and plain old unix pipes. The directives are simple XML processing instructions.
While not as clean as a DOM implementation, it has the benefit that you can add the tags you want and best of all, it's simple. I can edit it with a text editor and if desired, take the standard output of a command and insert it into the body of any given tag. (It's also possible to put perl code into a processing instruction, for example, to fetch information from a gopher server)
The generated feed is then a static file, you can run this as a cron process to update the feed once pr. hour or however often you'd like.