[eboyhan]

Most books on a kindle are represented by two files: one containing the document content (for example .azw, .pdf, etc), and one containing the document's "state" -- what page you are on, annotations, bookmarks, etc ( files such as .mbp, .pdr, etc).

Should these "state" files be referenced in addition to the content files as part of an add books operation? I did some tests with a number of pdf files, and if I included the .pdr files along with the .pdf files, I ended up with duplicate entries.

Yet there is apparently a fetch annotations command which references the .mbp files (I don't know about the .pdr's -- as I don't believe annotation of pdf's is supported on kindles)?

So under what circumstances if any does calibre deal with these "state" files; and what happens to any data fetched from them into the calibre environment when a send to device is done pushing a book back to the kindle? Clearly some state data is ephemeral, but other stuff is more permanent. It's not clear to me how coherency is or is not maintained.

Today, over in the Amazon kindle community a thread was started that says if a fetch annotations is done by calibre, and that document is subsequently sent back to the kindle, the annotations may be lost. Is this correct?

When a user wants to remove a book from the kindle, what happens depends on whether the book was gotten from Amazon or not. For Amazon books a copy of the book and all relevant "state" data is maintained on the Amazon archive. For non-Amazon books they are deleted absolutely -- unless some backup was provided by the user (such as by keeping a copy on Calibre). The claim is that calibre can restore the book's content, but not its "state". Is this correct?