I only have about a quarter of my epubs loaded into calibre so far, right now that's 2200 books. So as time goes on I'll be swinging a bigger and bigger mallet at CC. If anyone's going to break CC from sheer scale, it's probably going to be me
It's understood that short of writing some sort of plug-in that caches what's on an Android device, the initial sync is going to have to exchange metadata for all books on the phone. That's just reality. And trying to cache the contents of possible multiple Android devices that get connected to a single calibre instance would open up a whole 'nother can of worms.
The speed during the initial sync at connect time and especially during transfer aren't optimal (and the transfer speed seems to be slowing down even further as the number of books on the phone increases?!?!?!), but once I get my collection synced, the number of books to sync at a given time will drop to the level of whatever I add in an hour or so's work during the evening. I can set the phone by my bedside and let the sync run overnight, so long as there's no timeout that stops the process. So I'd say that if you can ensure that CC does an ack with calibre often enough that timeouts don't occur, that's by far the most important improvement.
Sorting and grouping taking 30 seconds doesn't bother me much; it's just that CC forgets that that's how I want my display to look and I have to go through the process again, and again.
I was thinking that once you built a display grouped, sorted etc. to a user's satisfaction, there would be some sort of staged data that reflected the display that could be referred to so the screen could be reloaded very fast. So if CC comes out of background it would just refer to the staged data to repaint the display rather than having to go back to the books to re-retrieve the metadata and re-group and re-sort it.
So far, it's when calibre has already broken the connection that I'd like to have a cancel function. Right now CC just hangs, and pressing return repeatedly does nothing. That's what I'd like to see fixed in one way or another.
One thing that has me nervous is that it's at the very end of the transfer process, after all books are transferred, that I see a message about updating metadata. Does that leave the risk of having several hours of transfers not being reflected on the screen if something happens 98% through a transfer? Would it be better to checkpoint the metadata after each n books, or does everything heal itself the next time a connect is done?
Thanks again. CC is a really cool product.