• Flazzle loading should be much faster as it loads modules on-demand. There is also a progress bar to show how much time is left before starting
• Added support for multiple archives. This is currently separated into “Flashcards” for tutorials, “HanDeDict” to allow automatic updates and “Other Dictionaries” (e.g. the Freedict, Japanese Edict etc) and I have plans to add the XDXF dictionaries. I think this is in general a good thing, however it does add latency to the downloads so I may create a listing with all archives combined to decrease download times in a future release
• Relaxed the licensing a bit as per my last post by removing the 15-day limit and making the basic dictionary and flashcards free for non-commercial use, and added the flashcard part back
• Vastly improved SQLite database speeds, sometimes by 30-50% when dealing with Unicode types
• Added manual bug/feature requests in the “Help” menu. This way you can request problems/features to be added to Flazzle (yes, I do always read and try to implement these wherever I can )
• Chinese input support is much improved with on-the-fly candidate conversion and the Chinese dictionaries have Pinyin combined with the “Word” column to allow easier definition lookup

What I am proposing to counter this problem is a relaxing of the licensing: possibly removing of the time limitation of 15 days in the unregistered version of Flazzle, allowing a free feature-limited use of Flazzle for non-commercial use.

I stand by my decision to eliminate the non-commercial standalone though, as it was totally unmaintainable as a separate code base and caused too many bugs. Flazzle is the product of approximately 10,000 hours of my time and though I want to be as open as possible, it would be extremely disappointing to me if it failed. Ultimately the goal of Flazzle is to make languages as accessible as possible whether paying for Flazzle, so my goal is to come to a compromise that both provides basic functionality for free and encourages businesses and people who want the extra functionality to upgrade.

Here is the proposed functionality and restrictions on the way Flazzle Free vs Pro operate:

This obviously isn’t final. What do you think? Would you pay for the Flazzle upgrade? Am I being too restrictive for the non-commercial version?

The other issue is of the flashcard program part. There’s been an (I believe legitimate) concern about removing a previously integral part of Flazzle. I’ve decided not to delay Flazzle any longer for now and add the old flashcard module back into Flazzle, but I do plan to rewrite this part after this latest release. The general inclination is towards a simpler, more universally usable flashcard interface.

As they say though, virtually no software is uncontroversial. You just have to learn from past experiences and try to correct the mistakes the second time around.

- David M, Flazzle.

• Fixed various encoding problems
• Relating to multilingual Windows versions and usernames:
• Config, temp and documents directories when there are non-ascii characters in the path
• Database connection problems where the database itself (i.e. SQLite and Firebird) doesn’t support Unicode characters in the path by changing the current directory and using relative rather than absolute paths
• Database connection messaging system encoding errors
• Most encoding “print” errors by replacing unprintable characters/auto-encoding Unicode types to UTF-8 when outputting to sys.stdout and sys.stderr which should eliminate this kind of error entirely in the future
• Flazzle now transliterates all non-latin and accented latin characters in filenames to basic Latin when saving to external files, which fixes uploading to FTP and the “common dialog failed with error 3002″ error when saving to disk
• Fixed various font fallback problems
• Updating of fonts in the config should be live now, for example changing the “Armenian” fonts should automatically update all armenian search results, editor etc fonts without restarting Flazzle
• Added font hashing, which decreases memory usage and reduces load times
• Added a few hundred more fonts to the font fallbacks for non-CJK languages
• Download preview windows now use font fallback systems
• Fixed font download dialog displayed multiple times for a given language after resetting a font fallback to default which was previously overridden when there was no default found. Only works for font download languages currently, e.g. Chinese, Japanese, Korean and classical Vietnamese
• Created a new, I hope more intuitive font download dialog, which now allows the choices between automatic, manual or choosing another installed font

• Changed the name of Flazzle Pro to be Flazzle WorldLearn for now, which should hopefully be a bit more obvious for new users. (I know Flazzle doesn’t truly have inputs for every language in the world, but it does have inputs for around ~200 as of Dec ‘06 which I don’t think can be much better summarized in two words. I’m open to suggestions, though )
• Various input-related changes:
• Onscreen keyboards are now favored over BySound systems to make it a bit easier for native speakers/first time users of Flazzle
• Fixed physical keyboard events not passed to onscreen keyboards. E.g., if the shift button is pressed, it updates the onscreen keyboard automatically now
• Added mapping of system keyboard keypresses to onscreen keyboards to allow speed typing if the user is reasonably familiar with the layout even if the layout is different to the physical keyboard
• Fixed rounding problems when magnifying onscreen keyboards
• Added new popup-window based accent selectors for languages written with a Latin script, e.g. French. For example, if someone presses “a”, a popup window shows with numbers assigned to each accent to allow speed typing, and pressing any characters other than a number will use hide the popup and use the original character. All characters outside the a-Z range are accessible using the “Aa” character selector button in the top left hand corner of input controls
• Added an “insert character” button to all input controls, which allows either insertion of accents which can’t be otherwise accessed from a menu, or using the character palette built into Flazzle

• Improved speed of matches in search interface
• Similar matches by ~6x and enabled automatic similar matches if no other matches found (it was too slow before and needed the “enter” key to be manually pressed)
• Wildcard matches by ~5x by only performing regular expressions on possible matches
• Added a new program to Flazzle. It’s currently called “Flazzle WorldEdit”, which allows using the inputs inside of Flazzle for basic wordprocessing and copy/pasting into other programs and has the eventual goal of replacing the buggy and unmaintainable “Flazzle Input” program

• Almost totally rewrote the networking code. It used to use the python ftplib/httplib which were great interfaces in that they were simple, but they didn’t have authenticated proxy support etc, so I changed it over to pycURL.
• It should work with most proxies now, and can be changed either globally in the config or at an individual connection level.
• It also adds support to other protocols including HTTPS, FTPS, FILE (and TFTP, but I don’t recommend using it because it has no security )
• The text-to-speech and font download interfaces now use the global settings proxy in Tools->Configuration
• Character palette changes
• Added displaying of Unicode codepoints, for example U+FFFF
• Added find-as-you-type to search
• Improved accuracy of search using a fallback system similar to the Flazzle search system: Exact, partial and similar (spelling errors) are tried in sequence
• Added new file format, as commented in my last post. The old SQLite format still exists because it’s more conservative on CPU and memory and is used internally by Flazzle’s character info, but the new format can often produce files that are 2-3x smaller than the old format. It uses 7zip LZMA by default, but the format allows different compression levels and formats, for example zip, bz2 and gzip, which may be added to a later version of Flazzle. You mileage may vary, but Jim Breen’s Edict, for example, is 1.5MB, down from 3.2MB
• Rewrote connect to server wizard, adding additional proxy options for RemoteUrl and providing different options for LocalSqlite, Local (Firebird), RemoteUrl (FTP/HTTP/FILE etc) and Remote (Flazzle Server SSH protocol)

• Added “unofficial” tags to all old version vocab lists to attempt to reduce confusion
• Fixed various remote problems
• Added much improved error handling
• Added ask before save when connecting directly using main interface
• Fixed irritating random hang bug relating to SNTP times not being retrieved
• Added transliteration to columns in column properties
• Reduced recursive populate problems for lists/trees in home/recycle bin and now only updates when changed
• Fixed import of vocab manually downloaded from a Flazzle FTP/HTTP etc archive for restrictive proxies. Still could use a bit of work, though
• Added stripe effects to preview windows and search results to give Flazzle a more unified look
• Removed tab-separated export support from “export other” wizard. It now has a dedicated wizard
• Almost totally rewrote the search interface
• Now uses the native HTML control for that platform, e.g. Internet Explorer or Webkit. Direct HTML is not currently supported for security reasons, though limited tags may be added in the future
• Character definition popups are handled by javascript
• Use of HTML for search results also allows possible reuse in a web version which could be released with Flazzle Server
• Fixed clearing selections

• Updated character palette to use indexes when filtering Japanese radicals, which should be around 2-3 times faster
• Fixed connections not displayed after adding in config
• Updated Flazzle’s version of ICU (International Components for Unicode) to version 6.0, which adds new transliteration schemes for Syriac, Armenian, Thaana and Georgian
• Removed certain parts of Flazzle which I couldn’t maintain/were too buggy:
• Removed all recursive subfolder code and changed to being flat-nested categories. This caused too many bugs and was taking too much time to maintain
• Removed the flashcard part of Flazzle. This may only be temporary, but it depends on whether there is any demand to continue development on the flashcard part. The general impression, though, was that it was too buggy, complicated and required too much intervention of the user. If it does get added back to Flazzle though, it will probably be rewritten with an emphasis on keeping it simple. If you liked it or have comments about this, please add a comment to this post!

The following changes have been made already:

• Hopefully fixed all of the database etc bugs relating to multi-lingual versions of Windows.
• Changed the default database to be SQLite instead of Firebird. This fixes the ~12000 limit on characters per column (which some English dictionaries use) and makes importing vocab as much as 3x faster. It only works with a single user at once, but that shouldn’t bother too many desktop users (Flazzle Server, if ever released, will still use Firebird!)
• Moved the character palette (which allows searching of characters by script/block/Chinese radical etc) to be under the “Characters” menu in the main interface, so more people should find it and is less cluttered to work with.
• A new, more minimalistic website design that should load much faster. Isn’t live at the time of this post (uses a similar theme to this blog though), because it needs to go through SVN (stability control) before release.
• The Non-Commercial version has been discontinued. This edition of Flazzle effectively meant that there were two separate code bases (the non-commercial built from the pro version), and it made both versions of Flazzle totally unmaintainable. It also reflects badly on my professional version because it just left both versions broken and buggy. It does not mean that I’m going to close the source to all parts of Flazzle, though. In particular, I plan to open the format specifications and allow conversions to/from Flazzle and other formats using a program released under a license such as the revised BSD. The non-commercial version will not be removed from the download archive (www.flazzle.net/files), but it will not be supported in any way. In addition, the pricing scheme of Flazzle Pro will be reduced to be $20 until the beta period is over, and it will cost nothing to upgrade to the stable version when released.
• I’ve got a G3 iBook off eBay and added Mac OS X Tiger support! OK, so this one is a bit optimistic, and definitely requires more work. Though it’s mostly working, there are still lots of gotchas and performance issues so it almost certainly will be released later than 1.6b (Screenshot attached.)

• And last but not least, a massive update to the search interface. It adds indexing which speeds up searches, “find-as-you-type” searching and has a “fallback” system:
• First, it tries for the exact matches.
• If it can’t find any matches, it searches for “partial” matches e.g. cat(ch).
• If it still can’t find any, it searches for “like” matches, and give spelling suggestions. From what I’ve seen so far, it works very well for spoken Japanese, and seems to work OK european languages too! This would have to be my favorite feature of Flazzle so far

The search interface, though it’s much improved, still has a long way to go before release IMHO. In particular, it just seems too empty, and performs too slowly on my iBook G3 900. So I plan to add a right pane that shows just results rather than like the results on the left in alphabetical order and just show one result at once to speed up the display of results (it can currently grind my 3200+ AMD Athlon 64 to it’s knees when there are hundreds of results, which happens quite often.)

I also would like to remove the irritating “too many files open” and “the operation completed successfully” errors messages (ironic, eh!) This ranks top of my annoyance list and seems to be caused by the wxHTMLWindow not closing the background image files when using ‘BODY BACKGROUND=”x.png”‘. So I may just rewrite that section with another widget to fix these problems.

Another criticism I hope to reduce (if not solve totally) is download sizes. I’m planning to separate the program into 5 different distributions: All languages, all other languages apart from CJK, and Chinese, Japanese and Korean respectively.

In addition, I’m planning to introduce a new compressed file format (bzip2, zip, lzma and gzip are all supported for the technical, though bzip2 is the default for packaging reasons even if it is ~15th larger than lzma), which potentially could be around 2/5 less space than the current SQLite format, in trade for processor speed. It also will allow “bundles” of flashcards so that the downloadable previews from the Flazzle archive should be less than half the download time.