[DNSB]

There are times when I feel that calibre should have used long numeric fields for both the directories and filenames to make it harder for people to grumble about the directory structure/filenaming conventions used. A random GUID lookalike for the author directory and the same with the calibre ID appended for the book directory? 99b1d595-2f03-40d0-89d6-01e7a5ed20d0/881d3ce3-458e-4145-928f-cd3fbffb76af-0012003 for example with the book filename being 881d3ce3-458e-4145-928f-cd3fbffb76af.epub. This would also solve those complaints about not being able to use extended character sets in the filenames.

I've had to deal with a few idiots who decided to re-arrange their calibre libraries and then "Help! My calibre library doesn't work!." One decided that he did not like the author/name/title (numbers)/ format and renamed/moved them to a author/ format which calibre was not happy with. At least, they did rename all the cover.jpg and metadata.ofp to booktitle.jpg and booktitle.opf so the structure was recoverable with some pain.

The other decided to move all the ebooks/cover images/metadata into a single directory for convenience. Sadly this ended up with one cover.jpg and one metadata.opf file in that single directory and calibre thinking there were no books in the library.

Both of them felt this was partly my fault for introducing them to calibre and partly the fault of calibre's author for not allowing them to use a random directory structure. Evidently calibre should emulate other programs that they use that allow to keep files where they want though moving those files can lead to "interesting" results.