Hi Mike
SENTENCE-BASED FC TESTING
You said you were thinking of having more sentence-based flash-carding and/or testing for a future version of Pleco. I've just been playing with setting up some sentence-based flashcards and seeing how the FC testing system handles them.
I was wondering - currently the limit for 'free answer' testing is 4 character headwords. A simple way to allow more sentence-based testing would be gained through extending the 'free answer' user interface to incorporate longer headwords? Perhaps with 8 or 12 characters. For example, on the character-writing 'free answer' screen, perhaps you could have a line of the 8 character boxes from left-to-right above the character entry box (and plus the 4 below, making 12 characters). Similarly for the pronunciation 'free answer' screen, 2 or 3 levels of the pinyin boxes (or maybe just smaller boxes).
I imagine though you'll be considering something even more flexible, allowing for even longer sentences to be written but just a thought for the short-term.
ALTERNATING TEST SELECTION
You also mentioned somewhere that you were considering making significant changes to flash-carding, especially with developing for the iPhone and the web (I hope I'm not putting words in your mouth again!). The current system is highly flexible and likely to satisfy most people that take the time to get their teeth stuck in to it. However, what I was wondering is, although you have 'alternating subject selection', why not have 'alternating testing' too? In other words, the system takes upon itself the creation of multiple score-files for each testing type of every card (E.g. stroke order score-file, free answer written score-file, free answer pronunciation score-file, multiple-choice reading [HEADWORD/MEANING, MEANING/HEADWORD] score-file, etc etc) and you can literally just jump in there as currently and have a whole bunch of different tests for different cards thrown at you and the system will gradually become aware of your weak understanding of one FC, your inability to write another properly, your misunderstanding of the tones of another, etc and use repetition-spacing or frequency adjusting algorithms to focus on these specific tests for these specific flashcards. You could have the character flashcards simply already set up in there for character-based testing. In summary (sorry for my long windedness!) to have not just an alternating subject selection but an across the board alternating test selection too (E.g. A multiple choice test for one card testing headword to definition, then write the stroke order for another, then write another's character, then another multiple choice giving the definition and asking for the headword, then to give the pronunciation of another with the tone, etc, etc and mix not just the user-selected flashcard words in but also pre-configured character FCs for the stroke orders and character meanings using the unicode info)? Such a system could be set up to be the fire-and-forget default configuration and still have plenty of the current back-end flexibility you already have tucked away somewhere.
The 'free answer' test is a big step towards doing this but doesn't go the whole hog and I'm not sure it distinguishes between the different testing or rather clumps it all into the one score-file.
Is this what you were thinking of doing?