Than you, Kristian.
The case where the books is exactly the same of course is irrelevant, it would never rise an error or an issue. That could arise when you change something into the book (if the filename is irrelevant, I guess metadata such as author, title and UUID).
It could be useful to write a FAQ telling people what they have to ensure is matching when importing annotation, and if something does not match how to edit both the epub and the xhtml file to get the both to match each other. An ID field would be the better thing, and maybe one can use the UUID from Calibre, but I wonder whether the epub gets edited with the UUID from Calibre (I knew Calibre do not touch epubs, as long as you do not install the modify epub plugin).
And let's talk this scenario: I have a Calibre library which gets messed up, I rebuild it from scratch, reimporting every book, in that case I guess the UUID would change, right?
And, eventually, what about letting the user decide to which book import the annotations while importing? Marvin could suggest the best match, but the user could scroll the library and choose another one...
My 2 cents, of course.
Have a nice week/end.
tiziano solignani, da Mac