Using the most recent build of Prince demo on Windows 7 and Linux (Fedora 16 kernel 3.3.4-3, linux generic 32-bit static build)

prince -o out.pdf http://...(URL to a server that returns an html document) works fine on windows.

Same command on linux fails with a segfault. Using --debug shows that the URL was resolved correctly and returned a valid resource (correct content type, charset etc) but right after that is the Mercury runtime segmentation violation. The 'address involved' seems to vary each invocation, e.g. 0xad000000, 0x9d000000, 0x91000000, 0xaf000000

On Fedora, I also tried creating an html file (wget -O file.html URL) and then running that through prince but that segfaults as well.

anyone have any ideas?

Might be a known bug with the new HTML parser affecting UTF-8 documents containing some invalid byte sequences. You can confirm this by trying it with the old parser, by passing "-i html4" on the command-line.

We have already fixed the bug for the 9.0 rev 2 maintenance release; updated packages are already available for some platforms, but not the generic Linux packages yet.

mikeday - Using the old parser indeed eliminated the segfault - thanks!

In the debug output now there is a warning about no glyphs for a number of characters (e.g. U+7e41). are these warnings related to the 'invalid byte sequences' you mentioned?

U+7E41 is a CJK character, so you might just need to install some Chinese and/or Japanese fonts on your server. By the way, we have now updated the Prince 9 packages to revision 2, which fixes the problem with the new HTML parser.