2.0.8 Beta 3 Released

mikelove

皇帝
Staff member
The latest 2.0.8 beta-test version (mostly just fixing bugs) is now available at:

http://www.pleco.com/beta/pd208b3.zip

And to test out the new E-C fulltext relevance-sorting feature, an updated version of the ABC C-E dictionary which enables that feature is available at:

http://www.pleco.com/beta/p2abc-100303.zip

Note that fulltext searches in this updated ABC will not work correctly with older versions of Pleco, though the rest of the database should still be functional.

INSTALLATION

To update on Palm, just install the updated application file to your handheld and it'll replace the regular 2.0.3 one.

On Windows Mobile, exit Pleco completely with the Quit command in the Dict menu, then connect your handheld to your desktop, go into its memory through My Computer and copy the updated Pleco application file into your \\Program Files\Pleco folder (My Computer, Mobile Device, My Windows Mobile-Based Device, Program Files, Pleco on XP, or on Vista, My Computer, (your handheld's name), /, Program Files, Pleco), replacing the version that's already there. As of 2.0.8, the Windows Mobile 2003 and Windows Mobile 5/6 versions have now been combined, so you can just install the one application regardless of which version of WM you're using.

PLEASE BACK UP YOUR FLASHCARD AND USER DICTIONARY DATABASES BEFORE TRYING THIS BETA - see http://www.pleco.com/manual/wminst.html#backrest for instructions on how to do that on WM and http://www.pleco.com/manual/palminst.html#backrest for instructions on Palm.

LIST OF CHANGES

BETA 3

Replaced the "last 10" field in Card Info with a "History" field on Windows Mobile (wasn't enough screen space do this on Palm OS), showing recent card answers / scores in reverse chronological order (newest first)

Also on Windows Mobile, switched to a better random number generator algorithm for shuffling / choosing frequency-based cards from flashcard decks, which should hopefully reduce the incidence of lots of unlikely cards coming up in a block

Changed the behavior of "Number of cards" "All" in Freq-adjusted mode; it now behaves exactly the same as a test with "Number of cards" set to "Endless" and "Don't repeat cards in session" checked in Advanced Settings / Cards, except with the addition of a total card counter (number after the /). So in other words, it will show you all of your cards as before, but the lower-score cards will be more likely to come up early in the session.

Fixed a bug that prevented the "if rep-spaced is late" label from showing up correctly in the Scoring tab of Advanced Settings on Windows Mobile.

Fixed a bug that prevented multi-syllable Pinyin searches from working correctly when the first Pinyin syllable was "shuang" or "zhuang".

Fixed a bug that could occasionally cause a repetition-spaced card to have its score / difficulty adjusted (or not adjusted) as if it were being reviewed early even if it actually wasn't.

Fixed a bug that prevented the correct / incorrect counts at the end of a session from accurately reflecting correct / incorrect answers that were canceled with the "Undo" command.

Fixed a bug with the "Force defns from dicts" feature (Advanced Settings / Display) that would cause the system to fail to update cards' definitions correctly when backtracking with the Back command.

Fixed a bug that could cause some dictionaries to be skipped over when cycling through dictionaries in the Character Info Compounds screen when the Unihan character reference database had been moved to a higher priority in "Manage Dicts."

BETA 2

Added "not correct in row" and "not incorrect in row" options to Card Filter so as to allow searching for every possible type of learned or unlearned card

Swapped the "No" and "Cancel" button functions in the repetition-spaced end message, since the German edition of Windows Mobile 6.5 has a bug that causes it to omit the Cancel button in those Yes/No/Cancel dialogs and exiting is a more important function than suspended review

Fixed a bug that caused weird characters to appear instead of newlines in the rep-spaced end message on Palm

Added a simple progress indicator for flashcard database updates on Windows Mobile

Added an error message when attempting to upgrade a "Locked" dictionary indicating that it needs to be unlocked first

Fixed a bug that caused right-to-left handwriting strokes to occasionally cause the system to flip between tabs in the input panel on Windows Mobile 6.5

Fixed a bug that prevented the Undo command from working correctly on flashcards answered incorrectly

Fixed a bug that caused the flashcard count to be incorrect when reviewing incorrect answers after a repetition-spaced session

Fixed a bug that could cause an erroneous repetition-spaced message to come up at the start of a Random or Fixed session

Set the "use in reader" flag to now be enabled by default in user dictionaries that have been upgraded to the new format

BETA 1

ADDITIONS

Introduced a brand new indexing system for user dictionaries; this should be much much faster than the old one, particularly in the document reader. It does require a database upgrade, however; you can upgrade your database by selecting it in Manage Dicts and tapping on "Configure" and then "Upgrade." (make sure to back up your dictionary database before upgrading) Un-upgraded dictionaries should continue to work for now, though a few other improvements (like the better Pinyin matching system below) won't apply to them.

Updated the flashcard database format just slightly to introduce two improvements: flashcard sessions containing lots of different categories / sub-categories should start significantly faster, thanks to an updated indexing system, and user dictionary entry headwords can now be edited without "Unlocking" them first thanks to a new index that makes it easy to search for / update flashcards that link to them. Flashcards will continue working perfectly well otherwise without this update, and neither change should prevent your flashcard database from being used with an earlier version of Pleco if you decide to revert to that. You can update your database through the "Flashcards" tab in Preferences.

Added an optional "Undo" button in flashcards; this will roll back any changes made to the card's score / review history. It only works when you're backtracking through cards; after tapping on the Back button once, one of the three answer buttons will be replaced with an Undo button; tap on this to revert the card to its score before you tested yourself on it. This will work with up to the last 15 cards tested.

Changed the algorithm for repetition-spaced card selection so that it will now prioritize cards by how far they are past their due date; this is done with a little bit of random variation in order to make sure that you don't keep seeing cards in the exact same order each time through. Also, when a repetition-spaced test ends you'll now be informed that you no longer have any cards due but offered the option to continue testing with cards that aren't due yet, with an additional option to not record scores for any of those cards (thus preserving / not extending their due dates); these will be sorted by their due date.

Overhauled the "Automatic" scoring system, though the old system is still available as an option ("Old Automatic.") Instead of Aggressiveness, all of the constants used by the system are now individually configurable through the "Tweak Parameters" option:

Difficulty change: the amount by which the card's difficulty is changed for an answer score of 1/2/3/4/5/6. (2 is the default incorrect score and 6 the default correct score if you haven't enabled detailed score buttons)

Initial score: the minimum score of a card after a correct answer; this can now be customized based on the answer score (4/5/6), which we've decided makes more sense than scaling it based on the card's difficulty.

Scale score inc %: this lets you multiply the amount by which a card's score will be increased after a correct answer by a set percentage based on the answer score (4/5/6); it's a multiplication, so 90% would mean that whatever score boost the algorithm generated would be multiplied by .9.

Scale score dec %: this lets you configure a percent applied to a card's score after an incorrect answer; the default 0 will just drop it to the minimum score (as before), but if you prefer to scale it rather than reduce it altogether you can do that using this option. As above, this is a multiplication applied to the score; if the score is 1000 and the "dec %" is 20 then the score would be 200 after a wrong answer.

Min / max score / difficulty let you configure minimum and maximum score / difficulty settings for cards.

Difficulty divisor is used in the formula for calculating a new card score after a correct answer. The card's difficulty (which defaults to 100) is divided by this amount, then multiplied by the card's score to get the new card score. This is an easy way to quickly make the algorithm more aggressive (by lowering this value) or less aggressive (by raising it) in how quickly it increases cards' scores.

Don't double-decrease unlearned difficulty: with this option enabled, if a card is "unlearned" (however you've defined that in the Cards tab of Advanced Settings), the difficulty will not be decreased again if you answer it incorrectly more than once in a row, reducing the likelihood of a brand new card being stuck in "difficulty hell" with an extremely low rate of score increase.

(on the main Scoring screen)

If rep-spaced is early: lets you specify how scores should be changed on repetition-spaced cards that are reviewed ahead of schedule (per the modified repetition-spacing algorithm above) and answered correctly. The default "don't change" is probably best in most cases - this will preserve the score exactly as is, so the interval between repetitions won't be altered. "change normally" increases the score exactly as the system would if you studied on time, "scale linearly" scales the score increase linearly by the time interval (so if a card was set to repeat in 6 days and you reviewed it after 3, you'd get 50% of the score increase), and "scale exponentially" squares that factor to make it smaller (so a card with a 6 day interval reviewed after 3 days would have its score increased by (50% * 50%) = 25% of the full amount).

If rep-spaced is late: works largely the same way, but "exponentially" is replaced by "logarithmically" since we're dealing with factors larger than 100%. With "scale logarithmically," if a 6-day card is only finally reviewed again after 12 days, instead of being scaled by 2x the score increase would be scaled by the square root of that (1.41x).

Added a new "Priority" feature to flashcards; this can be set individually through Card Info or on a group of cards through the Batch screen in Manage Flashcards. At present, priority affects exactly two things: which cards come up first in repetition-spaced sessions, and which cards are added to the pool of active cards first with "Manage Unlearned" enabled, though we may load additional functions onto it in future releases. Flashcard priority is now included in XML flashcard imports / exports, so it'll be preserved even after those. Priority is also displayable / searchable in the Manage Flashcards screen and displayable at the top of the flashcard session screen.

Along with priority, added two related options; one to force include a card in your "manage unlearned" pool, and another to exclude a card from all sessions (though this only applies to your current scorefile).

Added an "Overdue" option to the fields you can include at the top of the flashcard session screen; this will tell you how far overdue a repetition-spaced card is for review (or how long it'll be before it's due, if you're reviewing it early).

Added another new Manage Flashcards Batch option to add / subtract a random number of points (between 0 and a number you specify) to each card's score; this is an excellent way to randomly redistribute cards for repetition-spaced sessions if you find that you've got more cards due than you can currently handle.

Expanded the Card Info screen from one tab to two, to provide more space to see a card's definition and allow for the extra priority setting box / buttons to enable those force include / exclude options.

Expanded category selection screens (New Session / Manage Categories) on Windows Mobile to extend to the full screen height, useful on Windows Mobile 6.5 devices on which the height of individual items in those trees is greatly expanded.

Added options to the new flashcard session "Card Filter" tab to make it easier to set the filters to match your unlearned card check; cards can now be filtered by number of times correct / incorrect, and "last answer" has been replaced with a "correct in row / incorrect in row" option (just set that to "1" if you still want to check for the last answer only).

Added an "Unsafe Mode" option to the user dictionary and flashcard import screens; this skips several steps designed to keep a database file from becoming corrupted if your handheld crashes (or its battery runs out) in the middle of an import, but should also result in a big performance boost, so for large imports you might consider making a backup copy of your database file beforehand and then running the import with this option enabled.

Added new code on Windows Mobile to prevent too many files from being open at the same time - files will be closed and then reopened when needed when they get too numerous. This should significantly reduce the number of errors encounter when running Pleco on a device with lots of other applications open at the same time, or on a device (like some HTC models) with a very memory and file-intensive user interface shell / Today screen.

Added a "shuffle" option for "Repeat incorrect," which will randomize the order of your incorrect cards each time you review them.

Added an option in Test Setup to choose whether you want to be prompted each time you exit a session without saving or whether you'd prefer that the software always save / not save the session.

Merged the Windows Mobile 2003 and Windows Mobile 5/6 versions of Pleco into a single application, to get around the widespread problem of people accidentally installing both versions and ending up with the incorrect one overwriting the correct one.

Overhauled the Pinyin search system to no longer require spaces / apostrophes / tone numbers before syllables that begin with vowels. It will always do this in cases like "tiananmen" where there's no valid Pinyin parsing otherwise ("tiananmen," e.g., since "tia" isn't a Pinyin syllable and hence "tia nan men" is impossible), and with the new "Flexible syllable splits" option enabled (as it is by default), it'll work regardless; e.g. "xian" will now match both the single syllable "xian" and the mutli-syllable "Xi'an."

Added a new option to integrate full-text search into regular English-Chinese / Chinese-English searches. ("Integrate in E-C / C-E srch" option) With this enabled, after switching through all of your E-C / C-E dictionaries, the software will automatically try a full-text search of dictionaries in the opposite language.

Added an optimization to significantly speed up performance when only reading out a small subset of a dictionary entry; this will really make a difference with flashcards (almost doubles the speed of flashcard imports).

Added relevance sorting to fulltext English-to-Chinese searches; this can be disabled under Settings. Words are prioritized by whether they appear as the only word (or starting word) in a particular section of a dictionary definition, and words in definitions are prioritized over those in example sentences in some dictionaries. We've managed to do this in such a way that it has literally *no* impact on search speed, though it does require you to update your dictionary databases.

Added a "partial OK" option in Preferences (Results panel, next to Search Dicts); with this enabled, when tapping on the dictionary switch button, Pleco will stop on any dictionary that includes a matching entry for even just the first character of the search term, rather than skipping over those and switching to only the dictionaries with the longest matching length.

Improved the system for choosing single-character dictionary entries in the document reader, so that it's now more likely to pick the one with the most common pronunciation.

Changed the default tone colors to make the red / green a bit darker and more readable.

Added the ability for fulltext E-C searches to search for partial words, matching on the first word in its database that begins with the search text entered.

Improved search for ABC C-E dictionary links so that it will now factor in capitalization (needed to disambiguate entries in some cases).

Improved the sort order in which Pleco decides which flashcard database to open; if multiple databases are found, it'll choose the one with the earliest creation date, and if multiple databases are found with the same creation date it'll use the one that was most recently modified, which may help reduce the likelihood of accidentally opening up a backup copy of your database.

Added an error message on startup on Windows Mobile whenever more than one flashcard database is found, to further reduce the odds of using the wrong / a backup database accidentally.

Added the ability for some dictionaries (currently only need to use this with Adso) to force flashcards based on them to always store their definitions in the flashcard database (as custom cards), rather than linking to / updating definitions from the dictionary database; we instituted this because Adso has stopped using consistent ID codes for their dictionary database records.

Added an option for syllable-by-syllable audio to automatically reflect (simple) tone sandhi changes, 3rd-3rd etc; this is enabled by default for new installs, but should be manually enabled in Preferences for upgraders.

BUG FIXES

Fixed a bug that could cause flashcard statistics to return erroneous results when selecting a category with an extremely large number of child categories; in a few very extreme cases you'll just get an error message when running that statistics report, though.

Fixed a bug that could prevent Fixed sort order flashcard sessions from sorting their cards correctly, and that could prevent Manage Unlearned from correctly adding cards in the order created if that option was selected for it.

Fixed several bugs that could needlessly slow down the speed of user dictionaries displaying search results.

Fixed (we think) a bug that could occasionally cause some recent searches to be deleted from the search history.

Fixed a bug that could add some superfluous punctuation to ABC and ABCEC dictionary entries, though some of this fix won't apply until you update your dictionary databases.

Fixed a bug that in certain rare cases could cause some E-C dictionaries to be erroneously skipped over in searches even when they had results available that matched as well as other dictionaries that weren't skipped.

Fixed a bug on Palm OS that could cause crashes when invoking the Pop Defn / Copy to Input command on an empty part of a definition.

Fixed a bug on Palm OS that caused the "Instant" and "Auto-enter" handwriting options, when combined, to enter a new character in the Input Field after each new stroke was drawn.

Fixed a bug which made full-text English-to-Chinese searches switch to Chinese-to-English mode even when the text entered was still a valid English word (or the start of one).

Fixed a few bugs in "Chars Only" language auto-detect mode.

Fixed a few small memory leaks in some infrequently-used screens.

Fixed a major memory leak in full-text searching.

Fixed a bug that caused newly-introduced flashcard profile settings to have their values all default to zero; they'll now default to whatever the default value for a new profile is.

Fixed a bug that could confuse / scramble the characters in headwords in a few rare cases (traditional character mode, with a different number of characters inside the [] brackets than outside of them due to the presence of a /).
 

sfrrr

状元
After all that talk about flash cards (and how I don't use them), I've started using them again, so I may actually be able to contribute something to this discussion. Thanks, Mike

Sandra
 
Top