[sidd.artha]

I have been using Calibre for over 6 or 7 patches. And it has everything I wanted for my book colection and even a few things more. But everything stops when you end up using a test database of 3 books and move to the real thing. I have put a little over 2500 books, which is only part of the contents of my hard drive and everything went down with a bang.

Loading seems to go slower, although all it should do is read the internal database and let the rest for the time a users accesses information.

Once it loads it is huge in terms of memory. On a netbook with WinXP which has to fit in 1G of RAM you can't load something like a webbroswer and navigate as one would have to move to swap.

The autodetection is nice. But it takes only less than for ever. And there are some issues as it does not identify a book by its contents. It only tries to load the metadata stored infile. Say it does not find/read the metadata you get Unknown by Unknown and (!) you are prompted for duplicates. MD4 or MD5 are quick enough to help. Even a check of the filesize would show it's not a duplicate.

The last patch has shortened somehow the loading of books in the database, yet it will fail to read some files with an error.

And when it will fail with an error you have no way of knowing that the book WAS actually inserted in the database.

Which leads to another issue - for every tiny patch there is quite a file to be loaded. I'm sure at least the icon themes would not be that interesting to download over and over again. I believe that is a side effect of not having to pay for the actual bandwidth.

Also, there is an internal mechanism to access the net for upgrades, yet the upgrade has to be done with a browser. Why?

If there is the code for checking for upgrades, and the author has thrown in here everything but the kitchen sink (reader, converter, database) and there is a book reader which supports some HTML, yet for checking out the patched issues you get to fire the web browser. Which, as I said above, means a lot for a 1G machine. I'm sure that there are geeks out there with 1TB of RAM or even more, yet there are still out there laptops with 256MB or 512MB of RAM. They are quite good for having as book readers, as with my netbook, yet they could not possibly run an unoptimised mammoth like Calibre.

Although we're not talking about an internal database (say the bookmarks in Firefox), but a DB based on the filesystem it would be safe to assume that some changes are going to be made. The checking system is quite rudimentary. And slow. And it will not adapt to changes, you have to access some menu (can't say which as the metadata editor is stuck from some time, but I would like not to crash Calibre for fear of losing all my work). You access that command and than you sort the files for entries no lenght and erase those.

As pointed above Calibre can't find a duplicate even when it needs to know (for example if the Unknown by Unknown is the same book as the others with the same labels) so it could not do it for you either. Thus you need to use some duplicate file finder which, in turn, would make changes to the filesystem, which would disturb Calibre.

I have tried to make it slim by using or abusing the plugin system only to discover it is something... I don't know what, only not a plugin system. I see lists of plugins for handling different hardware ebook readers. I don't use any of them as I am missing the point - they are bulky, they have clunky interfaces, most of them give power to the firm that made them and not me, and, for the sake of turning an alleged 8000 pages I have to pay at least what I have paid for my netbook only to have less memory, more to fiddle and almost no other functionalty besides justify and an mp3 player. But that is another issue. My problem is that I can't unload those plugins. And others. What kind of a plugin is that that won't unload?

Getting metadata from the servers seems to work well, only that you have to close the book you are reading. It was very uncomfortable to do that as books that don't show up in major lists have to be entered by hand. Probably the author has assumed either that there are no books in the world that are not on amazon.com or that you should print each time you want to edit something. You can't do two things with the interface. So first you have to open the book. Than you have to enter the data in the meta editor. Than you have to kill the book reader. And only after that you can save the metadata. Fail to follow these precise steps in this particular order and you have to cancel, go back and retake the steps. And the book should not be opened as there is a need to rename the file which can't be postponed.

Saving the metadata for a 10 book database is quite ok. But saving the metadata for a 2500 book database means you have to wait about 30 minutes. At least that was my limit, after that I crash it. So far I found out that the data was written, but I always stay with the scare that everything would be lost. But as there is no cancel button (not even refresh so the metadata window will have ghosts of the other windows) I really can't see any other way.

There would be more things to report, but I already got tired of writing them down and you probably lost your patience to read this.

My conclusion: too good to be true. Wasting memory and a minute of loading time to manage a database of 10 books seems pointles. And when there's a large electronic bookcase the app becomes a pain. It rests the comfort of the convertor which I haven't tested so far, but which sports an impressive list of to/from formats. And the continuous list of supported news services which I hope they bring a good source of revenue to the author who could be able to make something of Calibre by the time it reaches version 1.0