Hi all,
I'm the author of Anki. Somebody pointed me to this thread. I appreciate all the interest!
As one of the recent posts mentioned, Anki is still definitely "version 0 software", and thus some things don't work as they should, or aren't documented properly. I've only been working on it for a few months, so please bear with me. :-)
Firstly, syncing fully supports multiple decks. It used to complain if you tried to sync a file with a different name to the one on the server, but a bunch of people complained that it was confusing, so I made it use the "default" deck if a deck only exists on one side. If you log in, then go to
http://anki.repose.cx/deck/add and create an empty deck using the same name as your local file, subsequent syncing attempts will sync based on the file name. I know this is fragile - and currently only the "default" deck is supported on the Keitai interface. It's one of the many things I'd like to improve in the future.
Next, generating reverse entries. I've been a bit hard-headed about this in the past, due to the tendency for people to want to practice Japanese -> English only, which is 'passive' learning and not the best way to learn. But I recognize that people's needs are different, and so the next release of Anki will include a check box to enable and disable reverse entries. As people have mentioned on earlier threads, importing from a text file doesn't do any reverse entry generation, so that's one workaround for now.
You're right that Anki is currently "2.5 sides" at the moment, too. Basically I would like to refactor the entire card handling system so that you are allowed an arbitrary number of "sides", and multiple cards can be generated based off a single vocabulary entry - so that if you discover a spelling mistake, you don't have to fix both cards like you have to do now. The only problem is that it requires a fair amount of work to change everything and make it general enough to be usable in a wide variety of situations, which is why I've been putting it off. Basically once implemented, you will have complete control over the number of cards that are generated from new input (you will be able to have a reverse entry, another entry for example sentences, etc). Tagging will also be supported which will function much like categories do. Because changing the card backend will make Anki a lot more flexible, I have been reluctant to add options like disabling generation of a reverse card - because in the new system, such options won't be necessary.
The MacOSX version has a number of rather annoying bugs at the moment, because I only ported the latest copy about a week ago. The biggest are the speed factor (it uses up spare CPU cycles effectively doing nothing), and the fact that using Ctrl+Space to toggle Japanese input actually inserts a space as well as changing the input method. These are bugs in the toolkit library, not Anki, and I'll need to see if the latest version of the library has addressed them.
Until recently I didn't have a Mac to test on, so I was relying on another person to build the Mac binaries for me. He stopped doing that months ago, so the Mac version of Anki fell quite behind. I finally got hold of a Mac and built the latest version, but I haven't had a chance to iron out all the bugs yet.
The problem downloading on MacOSX 10.3.9 is probably due to the lack of bzip DMG support. Perhaps there's a tool available online which will convert a bzipped DMG into a format earlier versions of OSX can read?
Please note that while the example decks are reasonably accurate, they probably do have a few mistakes, as they're based on the words in my personal deck that are also in the JLPT tests. I provide them as a way for people to see how to use Anki, and not so much as a study aid. I still think that entering things in yourself is the best way to learn, as you can use definitions which are meaningful to you, and the context you learnt the expression in helps your memory.
Megaqwerty: you said you were having crashes. Are you running Linux? Anki is quite sensitive to the qt version - please make sure you're using libqt4.2 and pyqt4.2. pyqt4.1 and 4.3 are known to cause crashes.
JimmySeal: Regarding settings being stored in a central location, your configuration is stored in your user data folder. You can passing --config /path/to/config to Anki to create and use a separate config file. This means you could create separate icons with different --config paths to effectively run Anki in different configurations should you wish. (One of my users uses it to run Anki off a flashdrive)
Anki has no trouble handling line breaks internally, but the sync protocol is line based at the moment. For now, just add <br> tags - I'll probably convert newlines to <br> tags in the future, in addition to adding explicit support for rich text editing.
Yorkii: I think in practice the numbers tend to be helpful, as you start to get a feel for how well you will remember things. I could add an option which hides the choices for those who'd rather not know, but since the scheduling time of the last card is shown up the top anyway, it doesn't seem like such a big deal.
If any of you tried the "web interface", you probably noticed the big red message up the top saying that it's quite minimal. My top priority is a javascript-enabled "study online" facility which is robust and easy to use. For many people installing software seems to be a big hurdle, and I want to make the benefits of spaced repetition available to complete beginners too. Of course, a new web interface will also benefit the rest of you who sometimes study at home, sometimes study on the go.
Vosmiura: those changes seem like a reasonable stopgap until I make things more flexible. I'll try and put out a new version soon.