[chaley]

Originally Posted by nqk

Last night (GMT+7) I got CC updated to 4.01 (beta I think, 'cause I can't see any alpha updates), I have the following comments and comparison with Calibre Box in regards of cloud services (Dropbox in particular).

First note: Calibre Box has a somewhat different purpose the CC's cloud connection. It is both the library organizer and the cloud downloader. CC separates those two functions. The consequence is that calibre box's behavior in the face of database changes etc is pushed to optimize startup because you start the cloud connect part every time you start the application. This is of course not true for CC. With CC you go to the cloud only when you want to download something into the local library.

- Picture/cover loading, database downloading: the two programs have similar/same rate.

Good

- Searching: To my surprise, CC was extremely slow. It had to scan 2200+ titles and returned the ouputs, while Calibre Box brought it on the fly.

Searching was the subject of much discussion when this project was started. Should the cloud connection use the calibre-like search syntax and semantics that CC and the content server connection use? Is searching even necessary given the hierarchical browsing? My compromise is the same as in CC base: support long-press searches that are database driven (and fast), and support the general calibre-style searches that require a scan. I have considered removing the general search mechanism, but came to the conclusion that there are times where it can really help such as with searches like "(#genre:vampire or #genrearanormal) and not #author:someone and #read:no"

Note that like in calibre, naked searches (searches without a column prefix) search all the columns. For example, searching for "foobar" will look in all of title, author, custom columns, comments, series, etc.

- Database checking: CC (re)download the database everytime. Of course the database file is changed when Calibre opens and closes, but it doesn't mean there is a need to redownload the file, I think. Manual download is better

I am not at all convinced. First, for me at least, starting and stopping calibre does not force CC to reload the database. I checked the dropbox logs and it did not upload the database. If you are seeing a download after calibre starts/stops then something in the database is really being changed.

If something in the db is really changed, then not downloading the db is simply asking for trouble. Seemingly small changes can make browsing produce the wrong answers, and even worse can prevent downloading covers or books. Making the user remember whether or not a change was "significant" is a recipe for bug reports like "CC failed to download my book".

I am willing to consider an option where CC informs you that the db has changed and allows you to accept or bypass the download. That at least tells the user that bets are off regarding accuracy, and gives me the answer to the inevitable bug reports.