Prince fails to generate PDF when the character "U+100010" is present in the html content for which the PDF is to be generated. The error log generated is

Mon Sep 20 17:36:27 2010: internal error: character out of range: U+100010

Is that a real character, or a broken encoding?

We do not know whether it is a real character or broken encoding. This is a character entered by some user of our system. I think prince should ignore such characters and generate a PDF for the remaining html content.

Right, we will change the next release to treat all characters the same. Since there are no glyphs for it, there will be a question mark in the PDF where it would otherwise appear, and a warning will still be issued, but Prince will not terminate and the PDF will be generated.

We managed to get that working by removing the character from the content itself.