Quote:
Originally Posted by BetterRed
The Shelf 'pages' calculator uses the book format file size to compute its value… that generates disk i/o
But that information is already available in the 'data' table of metadata.db. If the bookshelf 'page' calculator used that information I suspect it would speed up the initial scan by quite a lot… especially if that table is already loaded into calibre's in memory rendition of the database.
Attachment 220757
BR |
Calibre precisely does that.
And the format file size in 'pages' table is not used to compute the value, but as a guard: If you trigger the pages recount for this book, Calibre will compare the file size in the 'data' table to the size saved in the 'pages' table, and if their a equal, it skip the effective recount because it assume that the book is identical to the last time than the count was done.
This lead that if you request a recount of the entire library (and that you don't check the force option), the second scan will be even faster and will generates less disk i/o because unchanged books are skiped.