View Single Post
Old 06-19-2013, 11:25 PM   #104
DNSB
Bibliophagist
DNSB ought to be getting tired of karma fortunes by now.DNSB ought to be getting tired of karma fortunes by now.DNSB ought to be getting tired of karma fortunes by now.DNSB ought to be getting tired of karma fortunes by now.DNSB ought to be getting tired of karma fortunes by now.DNSB ought to be getting tired of karma fortunes by now.DNSB ought to be getting tired of karma fortunes by now.DNSB ought to be getting tired of karma fortunes by now.DNSB ought to be getting tired of karma fortunes by now.DNSB ought to be getting tired of karma fortunes by now.DNSB ought to be getting tired of karma fortunes by now.
 
DNSB's Avatar
 
Posts: 46,482
Karma: 169115146
Join Date: Jul 2010
Location: Vancouver
Device: Kobo Sage, Libra Colour, Lenovo M8 FHD, Paperwhite 4, Tolino epos
Quote:
Originally Posted by BWinmill View Post
Just a thought. I've heard that Be implemented metadata at the file system level. I believe that NTFS and ext3 both support it as well. (I forget whether is was mentioned specifically in the context of metadata, or that ACLs were implemented in a way sufficiently general to support metadata.) If this is the case, why couldn't the system separate the metadata from ePub and encode it in the file system as part of the file instead of a centralized database? This would allow for enough cross over to hopefully make everyone happy. (Kobo would still use the metadata approach, but at least it would be easier to manage without specialized tools.)
I suppose NTFS streams would work. I'm not sure if ext3 supports a similar metadata mechanism.

Quote:
Originally Posted by BWinmill View Post
JYet, I know there are issues with this approach. The biggest issue is cross platform support and support from file managers and other such utilities. I lived with the classic Mac OS for long enough to have a love/hate relationship with resource forks. But if the basics are there for the three major platforms (and, by extension, Android and iOS), why shouldn't developers start taking a serious stab at implementing interoperability.
For HFS/HFS+, my feelings would be more despise/hate. However since Kobo supports FAT32 on the exposed partitions, the only way I could see to handle metadata on a file by file basis would be to have the book processing routine generate a metadata file for each ebook (similar to the .opf files Calibre uses to store metadata) stored in a single directory similar to the image store currently implemented. IMO, we'd still need the database for speed in accessing information -- imagine doing a search by opening and reading 1000+ files.

Regards,
David
DNSB is offline   Reply With Quote