An option NOT to use authors and titles in directory and file names?

[Mekk]

It would be great, if calibre could be configured NOT to use author and title for directory and filenames, but use something persistend (like numerical id) instead.

Reasoning? Consequences of author/title spelling corrections when calibre library is version-controlled and/or synchronized – and I mean both explicit version control (I keep my library under mercurial) and automatic tools like Dropbox or other cloud drives.

Unfortunately those tools are usually bad at truly tracking the file over renames, even if they detect the fact, and - with the current calibre directory layout - fixing typo in the author name usually results in doubling space occupied by the repository, re-syncing whole (unchanged) file over the network, loosing the file changes history or similar problems.

I agree that the current convention is nice whenever one locates the book using file manager (insted of calibre), so I'd not recomment giving it up, but maybe calibre could make it possible to configure the way filenames are generated via some preference setting?

[Manichean]

See the sticky post on this.

[Mekk]

Originally Posted by Manichean

See the sticky post on this.

Both this post, and FAQ entry, carefully explain that there is no need to keep book metadata in the directory structure as it is better managed by calibre. I wholeheartedly agree…

[kovidgoyal]

The standaradisation of the directory structure allows various things to work well, for example the restore library tool. As such I am not going o change it. But you are welcome to make the change in running calibre fro source. It should be a very simply patch against database2.py to change the directory scheme.

[speakingtohe]

Both this post, and FAQ entry, carefully explain that there is no need to keep book metadata in the directory structure as it is better managed by calibre. I wholeheartedly agree…

There is no need but it does no harm.

I actually do find it useful on occasion if I find a book with missing or no formats in my current library. Easy to locate in a backup copy.

I occasionally scroll down the library file folder listing to check for author misspellings as well. Sometimes I spot things there that I miss in the tag browser. (I always change them using calibre not by changing the directory filename).

I can also imagine that the developer(s) would find it useful on occasion to see this info on an os level.

Longer file folder names can be a bit of a pain for those with an already complex (very long path before reaching calibre folder level) but it might be better to simplfy things on a higher level.


Helen

[Mekk]

Originally Posted by speakingtohe

There is no need but it does no harm.

As I said in my opening post - it does some harm. When you use file synchronization or version-control your library, every author name or book title change turns unexpectedly costly (in terms of network traffic or disk storage)

The problem is not in the name format, but in its strict dependency on metadata.

Longer file folder names can be a bit of a pain

In fact I happened to get "file name too long" errors when I tried moving my library deeper in my directory hierarchy, but there I exaggerated a bit in a book title, so it is not that important.

Well, Kovid said his opinion and I am to live with it.

[speakingtohe]

As I said in my opening post - it does some harm. When you use file synchronization or version-control your library, every author name or book title change turns unexpectedly costly (in terms of network traffic or disk storage)

Ahh I do not use dropbox as my library is large, so I am unaware of the implications.
I am not sure how changing the file structure would affect or improve this though.

If I change an author name globally from within calibre I simply get a new name on the directory. If I cahnge it on an individual book for authors with more than one book, I ge a new directory and author listing.

AFAIK there is no way to know without human intervention which authors/books are identical if you spell the names differently.

The dependancy on metadata for folder naming is restricted to author and title and the dependency of the title being associated with the author.

If for example, I have an author listed as Francis Paul Wilson and F. Paul Wilson whom I know are the same (human intervention ) How would them being listed in the folder structure as 8079 and 11136 and then being consolidated to 11136 make a difference from them being consolidated to F. Paul Wilson?

I am seriously interested in what using only numbers instead of author names would accomplish in database terms (or any other terms).

Helen

[theducks]

If you do what Project Gutenberg does and only assign a number to the title: Making a change to the title is only a DB change. Change the Author, same-o.

all would be needed is a re-write of metadata.opf for that number.
Way less net traffic. The backup of the metadata inside the book folder is there in case of a complete DB meltdown.

[Mekk]

Originally Posted by speakingtohe

If for example, I have an author listed as Francis Paul Wilson and F. Paul Wilson whom I know are the same (human intervention ) How would them being listed in the folder structure as 8079 and 11136 and then being consolidated to 11136 make a difference from them being consolidated to F. Paul Wilson?

First and aforemost, if you have *only* F. Paul Wilson, but decide to use his full name and update it to Francis Paul Wilson, then – with numeric directories – you have no on-disk changes at all, only metadata changes. Ditto for book titles (which I edit more often then authors)

In case of merge, if we insist on 1-1 between id and author, then copying turns necessary. But maybe one could just let have a few id's mapped to the same name - then again, no copying.