Quote:
Originally Posted by TechniSol
You could do that, but why bother? The method I outlined doesn't care about physical structure except when titles are added and it must record their paths which must be done in the dBASE already. If you wish to browse a "Shelf" set the dropdown lists to create the virtual shelf sorted by whatever criteria you like and voila. Think about how much this can speed up varying waits we experience now. All new titles can be dropped into one folder location and only that location need be checked each time you power cycle as deletions would go through the database as well. The ereader would only need to check for new books in that one location and not have to verify the database on every power cycle. It would move the new submission out of that directory and add it wherever it thought appropriate, or could even be set to preserve your directory/file structures. The actual physical locations are irrelevant, the file structure you see now is merely an abstract illusion represented to you by the file system.
|
The 'auto-create shelves by directory structure'-approach would cater to users who happen to already have a library well organized by file names and directory structure. Since re-organizing an existing library for metadata based library management is (1) a PITA and (2) not completely independent of how a specific device handles the tags I would consider this a valid use case worth some development effort.
The advantage compared to implementing a 'real' file browser is that, from the use case perspective, it would be reasonable to request a file browser to include file handling functionality. This is something that could NOT easily be implemented while keeping to a general metadata based paradigm for library handling - at least not on a device that takes as long as a KOBO to rebuild/update the database.
Assuming a completely and consistently tagged library there is, of course, no real need for a file browser style interface and I agree that in this scenario refining search, sorting and filtering options would be preferable. But this is a separate use case.