Hi all,
I have found the problem with image filenames not appearing in the OPF manifest. For each figure, I had one mediaobject containing two imageobjects: each imageobject had a role attribute to allow selection of the image for different output – one with role=”fo” and the other with role=”xhtml”. When I split each imageobject into its own mediaobject, the image filename appeared in the manifest. So when transforming to epub, the rule is: separate mediaobjects for each image in the XML. This is a trap: having multiple imageobjects in one mediaobject still produces valid XML. It also allows FO, XHTML and epub transformations without showing any errors. The images appear correctly in the resulting XHTML pages. I am sure you could create a valid epub like this too, but it would obviously be missing metadata. Maybe the epub transformation should fail if the DocBook markup does not satisfy the test of there being separate “mediaobject/imageobject/imagedata” for each image, because it would be missing metadata. Thanks, Dave Gardiner
