Feature Request: Customising metadata.calibre file content

[mornington]

It'd be nice to be able to customise what gets put in the metadata.calibre file which is sent to the device.

Obviously, some fields are required to be in there, but unless there's a pressing need why include custom columns and all of their details (like the formula used to generate them)? And why embed entire thumbnails in the json structure? Those things just seem to produce file bloat for no apparent purpose.

So, yeah, is it possible to allow the user to select which (non-core) fields are included in that file, and to switch off by default the sending of anything non-core to it, to reduce file size and hence transfer/sync time?

I've raised a ticket here: http://bugs.calibre-ebook.com/ticket/8001

[Manichean]

Why would anyone care on the user side of things? That file is just there for Calibre's internal purposes and doesn't get accessed in any way by any reading device that I know of.

[chaley]

As I said in the ticket, it would be very difficult to predict the effects of leaving random metadata fields out of that file, as it is used to build the full metadata information for books on the device. For this reason I am not in favor of such a change.

[mornington]

Originally Posted by chaley

As I said in the ticket, it would be very difficult to predict the effects of leaving random metadata fields out of that file, as it is used to build the full metadata information for books on the device. For this reason I am not in favor of such a change.

There's something which actually uses the binary thumbnails which are stored, json encoded, in metadata.calibre?

[chaley]

Originally Posted by mornington

There's something which actually uses the binary thumbnails which are stored, json encoded, in metadata.calibre?

They are used by the Sony device driver and by the device view.

[mornington]

Originally Posted by chaley

They are used by the Sony device driver and by the device view.

Do they actually *use* the json-encoded thumbnails or do they just write them to the metadata.calibre file and then read that data back again so they can write it out again later?

Are there any circumstances where the code needs a thumbnail from there which it can't just get from Calibre's or the device's own thumbnail cache, rather than by serialising/deserialising a binary file from the enormous json file?