We recently deployed the latest version of Aspose Words to our production environment and have had to roll it back to the older version we had been using as users were reporting that PDF documents they were generating were corrupt and just displaying the source XML in Acrobat Reader.
On investigation it appears that the software is getting into a cycle of creating a document successfully, but then creating a corrupt document of the very same type and then a successful one and then another corrupt one or sometimes the other way round with a corrupt one followed by a good one.
I can send you two examples of a document where the first generation was corrupt and then a regeneration of the same document was successful. Another regeneration proved corrupt and then the next successful and so on.
This is also not just occurring on the application’s front end but also for batch prints where the same behaviour has been detected resulting in a completely wasted print run at our outsourced printers. Once the Aspose Words software was rolled back to the older version this behaviour completely disappeared.