Well, the problem is that not many persons probably do use the migration functions export/import much, so it is not tested that much.
Further, I used calibre a long time on my old and trusty machine, starting with version 1.x or so. During all those years derelict and obsolete files seem to clutter the calibre environment as I noticed when inspecting and comparing old and new directories. So a fresh new start from time to time is never a bad thing IMHO.
Quote:
Originally Posted by kovidgoyal
Import does not re-create the full path to previous libraries, it only preserves the library name. This is because there is no guarantee that the full path is valid on the new computer.
|
Agree. Only on importing, the original full path-name with library name were shown (hence known), plus an input field so you could define the new destination. So I guess somehow the original path is remembered after all.
Quote:
Originally Posted by kovidgoyal
The important thing is that *all* the data is preserved.
|
I did a compare between the original directories and the new ones and noticed that all files are present on the new computer, minus file "metadata.opf". The latter was missing for all entries. No idea whether that file is important, can be recreated or is now obsolete.
Quote:
Originally Posted by kovidgoyal
I'm not sure how you ended up with paths as library names, but I cannot reproduce that.
|
In this case, by a rare coincident, the old source location and the new target destination *WHERE* the same, which *MIGHT* have helped to create the import confusion. Remember from my pictures: all libraries were shown with and without a path-name prefix, hence appeared twice.
At this moment calibre 64bits seems to work happily on the new ridiculous state-of-the-art machine using the old libraries. Even without those missing "metadata-opf" files.