JLDrill

Japanese Language Drill Program

Links:

• Mike Charlton's Homepage

Development Version

Latest Completed Development Version: jldrill-0.5.2

For details on development versions, see below.

If you’ve been following development of JLDrill for a while, you can see that stable releases are few an far between. I’m hoping to change that in the near future, but sometimes I’m lazy about updating this site. However, I actually do make updates to the code fairly frequently (usually every week).

Usually the development version is quite stable (it’s the one I use every day). So if you want something more cutting edge, or you would like to get involved in doing development on JLDrill you can always get the latest code from the JLDrill Git Repository.

I used to host development builds on my webserver. However, I am no longer doing so. If you wish to view the development documentation, you will have to download the code from GitHub and build it yourself.

After you get the development trees you will need to install the following gems:

• rake
• rspec
• rcov
• webgen

Once you have everything installed you will be able to build the gem using:

		rake gem

If you have all the debian deb-helper scripts installed, you can build everying with:

		rake update

Once you have done that, you should view the development documentation and TODO list for that version by looking at doc/index.html. It will also contain links to the user documentation for the specific version you build.

Development Versions

Development versions of JLDrill have a version number like 0.3.x where x is not 0. In Release versions, the number is 0. To understand development versions you have to understand my process a little bit.

I will give an example. Let’s say I released version 0.3.0. That is the “Release version” and you know it is because the last number is a 0. This is a stable release that I expect end users to be able to use. After I release this version, I will start to work on 0.3.1. This is a “Development version”. You know it’s development because the last number isn’t 0.

But while I’m working on it it isn’t completed yet. It might not even work properly. This is known as the “Current Development Version”. You can look at this web page to see what it is (see below). There is no reason to use a current development version unless you are modifying it.

When I am finished doing what I want to do for that development version, it will be complete. I will update this page to show what the “Latest Completed Development Version” is. I will also try to show what I’ve done to the version.

The Latest Completed Development Version is a bleeding edge version of JLDrill. I expect it to work. Some features might be incomplete and there might be bugs. But it should work. You can use it if you really need the new features, or if you want to join in on testing.

If you want to create a new release, please follow the New Release Procedures

Development Version History

Current Development Version: None Latest Completed Development Version: jldrill-0.5.2

Changes in jldrill-0.5.2

• Display statistics for forgotten items in a separate pane. These are basically the same statistics as the review set, except for the forgotten set.
• Show the appropriate examples when a question/answer is shown. Basically, if the question is English, show only English. If the question is Japanese, show only Japanese. When the answer is shown, show both.
• Support Ruby 1.9
• Chinese mode for drill. If your drill is chinese, you can select the chinese option in the Options window.
• CC-CEdict support. Add the CC-CEdict library (called cedict in the data files). If it is selected as the dictionary and chinese mode is turned on in the options, You will be able to search the chinese dictionary.
• If Chinese mode is turned on, show Chinese readings in the Kanji popup.
• Allow user to override datafiles by placing new versions in ~/jldrill
• Replace the Tanaka corpus with the Tatoeba database
• Allow loading of Chinese example sentences in the Tatoeba database (this is currently quite slow).
• If the chinese character is simplified, when displaying it in the kanji popup, search for it’s equivalent traditional character in the dictionary. If found, display that information.
• If the dictionary is loaded, do a dictionary search for the kanji character in the kanji popup. This gives a little bit more idea of the meaning, especially for Chinese mode.

Changes in jldrill-0.5.1

• Merge Context library with JLDrill code. Now there is just one deb file and gem.
• Allow switching between vocab and kanji popups without moving the mouse.
• Update Edict and Tanaka corpus from their source sites.
• Improve the Options UI
• Allow items to be forgotten if they have sat in the Review Set too long.
• Clean up Debian package so that lintian doesn’t complain any more.
• Create an official Gem package so that jldrill can be installed by simply typing “gem install jldrill”. This also installs the GTK libraries.

Changes in jldrill-0.4.2

• Allow items to be saved if only the comment has changed. When adding the items from the dictionary, don’t overwrite the comment.
• Load the Tanaka corpus and show examples of the current vocabulary being quized.
• Allow the user to search words in the dictionary by hovering the mouse over the word. This mode and the kanji lookup mode is switched by pressing the space bar
• Speed up dictionary searches. Especially for very short words the search could be slow. Now it is about the same for every word. It uses more memory now, though.
• Deinflect words in popup dictionary searches. This allows you to hover over sentences and find the correct word. Normal dictionary searches from the Add and Edit menus don’t deinflect yet.
• Bugfix where an error was given when you tried to save a file that wasn’t modified.
• Seach in the Add and Edit menu if you press return after editing the kanji.
• Create progress bars when loading any file. Also fix some concurrency problems. There are still a few pauses here and there, though. The progress bars themselves slow the load, unfortunately.
• Append File functionality was completely broken. It now works again.
• Fix crash when you press cancel when opening a file.
• Massive refactoring of the UI code.

Changes in jldrill-0.4.1

• Options to allow users to choose what kind of problems they want to drill in the Review Set. You can choose any or all of the 3 basic types of problems.

• Schedule each problem type separately. Originally the app would randomly select which type of problem to present when reviewing items in the Review set. Now each problem set is scheduled independently. But if one of them is wrong, they are all scheduled again. This has the effect that you review items more often and in some ways breaks the ordering of items in the review set (items that where reviewed recently with a different type of problem are more likely to be guessed correctly than their position would indicate). I will try to resolve this problem before 5.0.

• No longer schedule a kanji problem if the item has no kanji. This means that items with no kanji (like sentences in the grammar drill) are reviewed less often than items with kanji (like vocabulary). This appears to be less of a problem than it would seem.

• Create a maximum and minimum schedule for the item. Due to the way the user reviews an item may be reviewed right away or it may sit for a very long time. With this change the new schedule will never be less than the old schedule. Also if the item sits for a long time it might get a very long schedule. If the wait time is too long it is reduced to 125% of the previous schedule in calculations for the new schedule. See the Strategy document for more details.

• Change the multiplication factor for the new schedule. Originally it always multiplied the wait time by 2.0. Now it multiplies it by a factor of 1.0 to 2.0. New items that have waited less than a day will be multiplied by 2.0. Old items that have waited 180 days or more are multiplied by 1.0. Other items are somewhere in between. See the Strategy document for more details.

• Properly reset the Review Mode button when a new file is loaded.

• Fix a crash when editing vocabulary. I suspect this is actually a bug in GTK, but I added a work around anyway.

• Allow the user to specify a drill file to load on the command line.

• Add an option to specify the dictionary to load. This means that different drills can use different dictionaries. Also add an option to automatically load the dictionary when the drill is loaded.

• Move the dictionary loading progress bar out of its own window.

• Reorder the items in the new set when their order is changed in the All Vocabulary window.

Changes in jldrill-0.3.5

• Desktop entry added for GNOME. This will result in an entry in the GNOME menu. Also updated the icon to SVG.
• Reset Difficulty over time. If the user goes for a certain period of time without seeing an item and subsequently gets it right, reset the difficulty to match the time the passed.
• Allow the user to cut and paste items in the vocabulary list with C-X and C-V
• Add a timer for each question that causes the background to change red when the user spends more time thinking about the item than they did last time.
• Create a Debian package.
• Simply scheduling strategy so that items are sorted by the percentage of time passed in their potential schedule. This simplifies things dramatically, removes the concept of getting behind and also improves review performance.
• Number items properly in the kanji drill
• Add a Learn command that promotes the item into the review set immediately.
• Automatically adjust pane sizes so that the questions and answers fit as much as possible.
• Allow users to add a simple image to each item in their drill using an image: tag. Currently no documentation for this feature because it is likely to change.
• Simplify the statistics window slightly.
• Various test refactoring
• Allow jdrill to be run directly from a source repository by running the AppRun script.
• Configure the source repository so that it looks like a ROX Application Directory. Admittedly few people are likely to be using ROX-filer, but it’s cool for those that do.

Changes in jldrill-0.3.4

• Add description for 々 in the kanji popup
• Pressing Cancel when quitting used to leave you hung with no window. Fixed.
• Fixed a crash, but didn’t record the reason for the crash when I fixed it.
• If you have a kanji question and edit it to take out the kanji, the question in the main view used to be blank. Now it creates a new meaning problem.
• You can now load the dictionary from all windows except the Show All Vocabulary window.
• Improved the workflow for the add and edit windows (see documentation for details)
• If the user has seen the answer for a question, display it after the vocabulary has been updated.
• I now renumber the item positions when resetting the drill. This is to offset a very old bug that led to some invalid files.
• Editing items was sometimes slow due to searching more times than necessary.
• Considerable amount of code refactoring
• Make the all vocab list a non-modal dialog and it follow what’s going on.
• Add a word search in tables by pressing Ctrl-F
• Considerable speedup for appending files
• Allow editing, deleting and previewing items from the all vocabulary list
• Allow previewing from edit and add windows
• Allow deleting items from the main window
• Allow the user to change the order of items in the quiz in the all vocabulary list
• Allow the user to create a new drill.

Changes in jldrill-0.3.3

• Improved development documentation. Opening the index.html file in the doc directory gets you started.
• Some refactoring with Strategy and Item Status.
• Reset now drills the first item in the quiz. This means that when you reset the drill, there will always be one item in the level 1 working set. But this is better than keeping the last problem around, or not having a problem when you reset the drill (both of which are confusing).
• Only start the countdown for New Item mode after you have seen 10 items. I did this because I found that when I’m on top of things my review for the first 10 items is always 100%. Possibly this means that I need to back-off the difficulty for items that have already been memorized. More research needs to be done.
• Properly reset the scheduled time in items when the quiz is reset. Also reset each item when it is first introduced in case we have old quizzes that contain the scheduled time.
• Modify the stats window to show the accuracy of items by original duration rather than number of consecutive correct answers. This gives a better indication of whether or not the back-off is working properly (for difficult items I level 5 item may have only been around 2 or 3 days).

Changes in jldrill-0.3.2

• Dramatically speed up reference dictionary loading time.
• Reduce memory usage to some degree.
• Automatically load kanji information on startup.
• Fix bug where app could hang searching for an empty vocabulary.
• Clear the search box after an item has been added.
• Do a reference check for the current problem once the dictionary has been loaded.
• Automatically do a search in the add and edit windows once the dictionary has been loaded.

Changes in jldrill-0.3.1

• Add Rakefile tasks to build releases
  • publish – Publish the web page to Rubyforge
  • release – Create a release directory for the current version
  • update – Run rcov, web, and release
• Allow reading of Edict files in either EUC or UTF8 encoding
• Update the EDICT dictionary to the latest version

site info

© 2005-2011 Mike Charlton | Generated by webgen | Design by Andreas Viklund modified by Mike Charlton.