> At the risk of confusing you, incomplete 'URLs' in
> the scan rules are not relative URLs.  They are
> partial patterns matched against all (absolute)
> URLs.

Initially, I hesitated to include the above paragraph I wrote earlier because
it wasn't clear it was relevant, but more importantly, looking at it again,
it's also confusing, ironically enough.  I'll try again:

Some scan rules look a lot like relative URLs, but they aren't.  They're
actually merely patterns that match parts of URLs.

It's not clear that you didn't know that, but it seemed worth mentioning in
case there was any confusion.