View Single Post
Old 04-11-2013, 11:23 PM   #3
BetterRed
null operator (he/him)
BetterRed ought to be getting tired of karma fortunes by now.BetterRed ought to be getting tired of karma fortunes by now.BetterRed ought to be getting tired of karma fortunes by now.BetterRed ought to be getting tired of karma fortunes by now.BetterRed ought to be getting tired of karma fortunes by now.BetterRed ought to be getting tired of karma fortunes by now.BetterRed ought to be getting tired of karma fortunes by now.BetterRed ought to be getting tired of karma fortunes by now.BetterRed ought to be getting tired of karma fortunes by now.BetterRed ought to be getting tired of karma fortunes by now.BetterRed ought to be getting tired of karma fortunes by now.
 
Posts: 20,558
Karma: 26954694
Join Date: Mar 2012
Location: Sydney Australia
Device: none
Quote:
Originally Posted by Adoby View Post
Yes it COULD. But I don't think it should. It is safer to first copy all the files, then if successful, delete the original files. Then all or none are moved. I suspect this is intentional.

I prefer safety before speed, when it comes to how calibre manipulates files.

If you need speed, invest in faster disks and more memory for disk caches.
Adoby,

A move within a POSIX compliant partition is a heck of a lot safer than a copy, because it involves a lot less I/O.

A move is a copy and delete of an inode within the directory structure rather than copying and deleting whole files. Many file systems (NTFS, HFS+ etc) give the directory structure extra safeguards to reduce the chance of corruption - such as pre-allocating disk space, using contiguous clusters etc.

And, on a POSIX compliant file system Calibre in fact does do moves (copy and delete inodes) when a book or author is renamed within a Library (directory). So why not do the same when a book is moved between two Libraries (directories) in the same partition.

IMO GC has made a valid suggestion - that is in fact safer.

BR
BetterRed is online now   Reply With Quote