This is geared more towards Rdl I guess but others may know or be able to help with this. I apologize if I'm going about this the wrong way but I found a bug in the original sourceforge par2cmdline source that seems to also be in the modified build that is bundled with the altbinz application. The bug is covered in my post over on the quickpar forum at
http://www.quickpar.org.uk/forum/viewtopic.php?pid=5674 In short is has to do with creating par2 recovery files when files in the data set are very large. I don't know the exact cutoff limit, seems to be 4gb, but the original code seemed to use a long as the file size pointer instead of unsigned int64 like was intended in the code. The same bug seems to have an impact on the ability to verify par2 files (basing this on my tests with the original code not the modified) where it locks up in an infinite loop after reading all the data set files. From one of the other posts in the forum pointing towards the multi-threaded build that version has a similar fix like I tried and seems to support unicode but wants intel related stuff and an extra dll boo.
-- edit --
Sorry I was mistaken the post I saw about a multi-threaded version was on the sourceforge forum
http://sourceforge.net/forum/message.php?msg_id=4531616
