kdrill v5.00
  Primary "patch":
  	Started change to linked-list format.
	Converted just about all routines to expect a pointer-to-entry,
	instead of just an index into "translations".

  Reads in "edict"!!!
  
 Fixed stupid sprintf syntax error in MakeLog()

 Now C++ friendly! Can be compiled with g++. Hopefully, also with
 other C++ compilers, in case you don't have an ansi-c compiler, but happen
 to have a C++ compiler.  

kdrill v5.1
   Arrggg. Somehow, I trashed the dictionary reading methods, so the
   frequency index (or any others) never got read into memory.
   Fixed.

   And since I'm forced to make another release prematurely, I'm adding a
   bunch of forward stuff that was ment for v6 & the motif version:

   Slightly smaller memory usage (saving 40k, out of 5 megs, maybe :-)
   
   Added "4-corner" kanji search method! See docs for kanjidic for
   information on how to use. It's really rather complicated.

   Note: use kterm in EUC mode to view the kanjidic.doc file.
      Change to EUC mode by control-middle-button on kterm

kdrill v5.1.1
    Fixed stupid trigraph warning

kdrill v5.2
    Added blank for search "#x" window, when the "kanji" being shown
    is just a dummy to denote there IS no kanji!

    Fixed "in order" quizzing, that I had broken.

    Added a few more comments in Imakefile, and INSTALL

    Did some cleanup of annoying warnings.

    Restored the "compilable by C++" feature. (was missing prototypes)


kdrill v5.2.1
    Added a few more comments here and there

kdrill v5.3
    Read edict in a bit more sensibly. Split into kana and kanji
      readings, when available.
    Make search window wider to accommodate above.

kdrill v5.3.1
    Fixed dumb bug in errorlog replay... it picked random chars,
    instead of one you definately missed!

kdrill v5.3.2
    Fixed dumb "bug" in resource file. KDrill.ad did not have the
    new kdictfile/edictfile resources in it.

    Officially merged in "kana suppliment", in "makedic" directory.

kdrill v5.4
    Added "blank" entry, that is an alias to the true blank, in
    the four-corner kanji search popup.

    Enabled romaji printing and input!

kdrill v5.4.1
    Fixed omission of "wa/wo" in romaji inputconversion.

    Added transparent matching of kanatana->hiragana, in kana search.
    Hiragana in input, will match katakana in dictionary.

    Found and fixed source of annoying "null child" warning.

    Made the search window respect the kana/romaji toggle
      (although NOT in the input window, due to ambiguity)

    Added a hiragana/katakana toggle to the kana search window, for
    stupid people like me who haven't learned their katakana yet!
    [note that while you can click on katakana, it outputs hiragana.
     Which is "okay", since we have hira/kata insensitive search anyway. ]

    Added a mirror of the kana input window, in the kana popup!
    (about time)
    
5.4.1.1
	Minor stupid bug fixed... sometimes the "kanji24/16" font aliases
	are not present.. but the fonts we need are.
	I finally stopped being lazy, and made kdrill do a general
	request for jis fonts. Look at "KDrill.ad", and search for
	'kanjifont' for more details.

5.4.2.1
	more stupid bugs, plus I lost track of things for a while.
	Fixed -gradelevel didnt' work. somehow, GetOptions() got
	messed around with.

5.4.2.2
	improved makedic dictionaries. eventually, add thingie

	added warning for null or shortened usefile
	except 'usefile' doesn't seem to WORK ANY MORE?!?!?!?

	{should also warn for ignored kanjidic lines, so if corrupted,
	  warn?}

5.5
	FINALLY added paste drop-target in the search window!
	This adds nothing to kdrill as a quizzing tool... but
	makes it indispensable when trying to read stuff online
	with "kterm", and you're wondering what the heck a particular
	kanji means.

	Changed to always use bzero, instead of trying to please everyone.
	Happily, Xfuncs.h #defines bzero if missing.
	Improved the font lookup yet more, so that it cannot accidentally
	use the old "jis-201" standard.

5.5.1
	silly me. I forgot I had more than one way to view kanji.
	kterm does cut-n-paste encoded.
	NETSCAPE just does raw shift-JIS.
	So added "raw" decoding for drag-n-drop search

5.5.2
	minor corrections to README and manpage, about the monash ftp
	site.

	re-fixed gradelevel subsubwindow position. (wierd)

5.5.3
	Yowch. Nasty bugs in kana input by romaji.
	It was not properly translating romaji to kana

5.6	WHEEeeeee.. lotsachanges.

	Added button to force kana search to be at BEGINNING of a word.

	'zi' input romaji was not correct.

	Changed romaji input, so "-" does "-" in kana, but
	"'" does small-tsu.(previously, "-" was small-tsu)
	(May have fixed this earlier, actually)

	Patched REALLY REALLY bug in romaji printing, to do with trailing 'n'
	That is to say, 'n' by itself(no following vowel) was never printed.

	made search window update immediately when romaji/kana toggled

	Note added about buggy netscape cut-n-paste

	Improved makedic/README by documenting input format a little

	Fixed possble widget position bug, in option window?

	Finally fixed usefile to work again properly. I hope.

	Allow "on-the-fly" changes to in-memory usefile (via toggle button)

	AUTO-SAVE both usefile and logfile on exit

	Improved NOTES file somewhat

5.6.1
	Small doc improvements

5.6.2	Fix romaji input to be coded more redably.
	Also fix having to always type "n ", when sometimes it should
	be obvious to convert. Like "konnichiwa"

5.7	 (Happy birthday, Jenni! Dec 21st, 1998)

	Finally made popup windows handle WM_DELETE message from window
	manager
	(although slight bug still in the kanji search window.
	 I can't figure out why?!!
	 It SEEMS the same as the rest of the stuff...???)

	Added help description to four-corner window.

	Finally fix it so you can change DICTLOCATION in Imakefile, and
	KDrill.ad will AUTOMATICALLY get the right values!

	Cleaned up OPTIONS window layout, once and for all.
	There are now no fixed sizes in optionsform, so it is friendlier
	to really-large-font choices. Hopefully, i will never have to
	reorganize it again!

	Added rya,ryu, ryo to romaji input search

	Finally put kanji search stuffs, IN "kanjisearch.c", instead of
	  kanasearch.c

	Added MultiList popup for multiple match on searchstuff.


5.7.1
	Silly me, forgot to enable multiple matches on kanji search.
	Now when you cut-n-paste a single kanji, it will show all matches
	for that kanji in known phrases. (Assuming you have edict loaded)
5.7.2
	After more use of kanji cut-n-paste with string search enabled...
	I suddenly found that handling multiple kanji is very useful.
	So now, after long last... you can cut and paste a string of
	MULTIPLE kanji (up to three kanji in length) and kdrill
	will try to find a match. Hurrah!



5.8	[ This is huge. I'm tempted to make it kdrill 6, but...]
	[ I think that will wait until I support the new kanjidic format ]

	[5.8beta1]
	Added SKIP search capability !!

	Patched fourcorner lookup, to NOT use all lines in edict, if
	"0000" is generated as Q-index.
	(initialized Qindex to -1)

	Split out widget creation from "search.c", to "searchwidgets.c"

	Fixed up the four-corner display, so that the corner selected
	is REVERSED, instead of greyed out. Looks more logical now, hopefully.

	Made kanasearch window have button labeled "search" instead of
	paragraph sign.

	Finally made the Search window, be a top-level window in its own
	right.
	This means it can be iconified, or NOT iconified, all by itself!
	(So if all you're doing is searching, you dont have to have the
	 main window in full view any more)
	 [except still bugs in this.. sigh]

	Fixed romaji toggle to actually update display IMMEDIATELY

	Finally added SAVE of options, to users's home directory!!
	   $HOME/.kdrill

	Changed cursor in the "kana input" window, to be a proper text cursor.

	[-gradelevel might not have been DISPLAYing, even though worked.
	 I'm fairly sure I fixed that, too]

	Make status say "You missed this one before", (if missed) even if it
	wasn't explicitly picked as part of a "go back and redo" loop.

	FIXED problem with closing subwindows with WM_DELETE !!!
	Now ALL windows should be closable by window manager, safely.

	beta3:
		Increased readsize of line reads
		Added new way to calculate wrong kanji frequency.
		
		Fixed not saving bug in beta1, and also fixed
		-questionmode flag.
		
		Increased limit on multikanji, to 200  (MAXTRANS)

	beta4:
		truncated romaji strings that were too long. buffer
		overrun. Bad programmer.

	Release time!!

	After days of fiddling around, finally found a workaround for
	apparently broken X libraries in some versions of Redhat and
	FreeBSD. Opening the SKIP window would crash the program.
	Lesson: always name your widgets uniquely. Never use
		XtCreateWidget("1", ....);

5.9	Oct 27, 1999

	Simplified getline() to just use fgets().
	  If your system stdio is not 8-bit-clean, this will cause
	  problems !!

	Increased MAXTRANSLATIONSALLOWED. edict is unbelievably HUGE!!

	Added "gya","gyu","gyo"

	Moved "TallyWrong()" to game.c where it belongs

	Made numberincorrect count be INDEPENDANT of usefile, grade, etc.
	  Removed from CountKanji()
	  (To fix NEGATIVE counts for kanji missed !)

5.9.1
	Allowed saving of usefile, even if too few entries.

5.9.2
	Fixed kana search bug with "[..." option.
	Search for XX at start of word, will no longer match
	   ABC(XXX)
	Updated the INSTALL file, to hopefully be more usefull
5.9.3
	Allowed searching with () chars, for romaji/kana search. eg
		ki(ru)
	
5.9.4
	Updated readfile.c so that if it cannot load a dictionary, it
	prints the NAME of it!
	Plus did minor cleanup of loadusefile printfs

5.9.5
	Finally added the facility to easily view and remove
	lines from your usefile! Clicking on "show" in the search
	window will bring up the list.

5.9.6
	Added extra cut-n-pasted decoding in searchwidgets.c, to handle
	current Netscape.

	[cheated, and re-released 5.9.6 with updated man-page]

5.9.7
	Added 'C', for SUPERCHEAT: pop up search window with
	full correct kanji info. Useful for learning kanji.

	Allowed button3 to also trigger search window

	Added mya/myu/myo/nya/nyu/nyo

5.9.8
	Made '+' accelerator work properly, for filtering out
	grade levels. Plus fixed status message for '6'

	Allowed user to erase wrong count, with backspace or del key


5.9.9
	Fixed a lot of compiler warnings. And also fixed some actual
	array bounds type errors, thanks to compiler warnings sent
	in by Falk Hueffner
	Fixed hiragana/katakana toggle for "--" in kanainput window
	Added status label mirror to search window
	Fixed initial non-blank string in kanainput window
	Put in a redundant check on options window coordinates, for
	those people who run broken window managers.
	Fixed the default "first-time" clean install run, to display
	  Kanji-> English question, not kana->kana!
	Similarly, fixed fallback defaults to be -nousefile
	
5.9.9d	Fixed stupid Imakefile bug for InstallProgram, so
	"make install" will work again.
	Added kdrill.xpm
	Added manpages for makedic/makeedict
	Fixed default KDrill.ad generation for initial question mode.
	  (followup on 5.9.9 fix for that)
	  
5.10	Added "search for similar kanji" button ("match")
	Made match at start button ("[...") work for english as well.


5.10.1	fixed initlog to allow the logfile to be a different name

5.11	Added "learn" popup. Now you can have a nice SMALL window to
	memorize new chars with, instead of the bulky main or search
	window

5.11.1	Fixed missing break in ChangeMode()
	Augmented mainwindow accelerator: can now change both
	question, and answer display for E/K/M via keyboard.
	In the process, moved changequestion from options.c to widgets.c,
	  where ChangeMode() is
	Tweaked offsets in search window a teeny bit
	Moved some stuff into utils.c
	Added GetWidgetHexval(), and cleaned up DoFind significantly.
	   Got rid of nasty FindIndex() hack entirely, and made both
	   #x and Ux search inputs behave the same. Dont need 0x now.

5.11.2	Fixed Learn window, so that "Next" is different from
	"Next(Random)"
	While I was there, added keyboard accelerators 'n' and 'N'
	for learn window. And removed useless DoLearn() func.

5.11.3	Added Timer option, at long last. Yaaay!
	 'T' turns it on/off. See option popup for length adjustment.
	Unlink usefile and/or kdrill.log if zero contents
	Cleaned up function names GetXtString, etc, to be
	GetXtrmString: indicates "resource" related, rather than
	 widget related

5.11.4	Added getMultiCount(), so that I could...
	allow external routines to chose whether to force ShowMulti visible
	Reworked main-page widget holder, from formWidget to panedWidget.
	This forces all three top sections to fill the page, which makes it
	look nicer when odd fonts are used for the buttons.
	Unfortunately, makes it look more boring :-(
	Updated the ancient INSTALL file to be more readable, and
	added teeny bit of info.

5.11.5	patched BADCACHESIZE overflow bug
	fixed patchlevel.h dependancy in Imakefile, I hope.

5.11.6	Added enforced check for too few kanji when enabling usefule
	by buttonclick.
	Added hya/bya/pya to romaji input

5.11.7	Fixed bug in matchromaji() that was giving bogus matches too soon.
	(eg: 't' got matched to 'ti')
	Moved certain SetWidget/GetWidget functions into utils.c where
	they belong
	Added usefile toggle in multikanji search display.
	  Unfortunately, 'u' does not display correctly.

	Did a bunch of restructuring to work around the bug in Xaw.6+,
	where they BROKE auto-unhighlight behaviour for commandWidgets.
	Grrrr...
	Side effect: 'cheat' is now a toggle highlight for the current question.

5.11.8
	Fixed bug(?) in cut-n-paste. Uses doFunkyGrab() again.
	Cleaned up a bunch of compiler warnings from gcc -Wall
	Fixed romaji input. AGAIN. (From 5.11.6, I think)
	 Romaji conversion/input really needs de-uglification.
	 At least I added some more comments for next time around.
	
5.11.9
	More romajiinput tweaks: fixed pointless warning.
	Also, somehow fixed cut-n-paste bug in 5.11.8 package??

5.12
	Arrg.. runaway version numbers.
	Fixed buffer overflow in doJISConvert, that made coredumps
	for long cut-n-paste strings.
	Plus, fixed Yet Another 'cheat' bug, where if you cheat, the
	 button stays on after the NEXT choice.
	 (no-one reported this one. Odd...)
	Fixed uninitialized 'strparse' var in Handle_searchPasted()


5.12.1
	Added [return] key accelerator for stroke count search.
	Added "dzu" to convert.c
	Reformatted search window, and added "active only" filter
	Fixed stupid bug in matchengfirst so "Match initial" works with english
	Added "usefile" toggle to 'learn' window.
	Added trivial #ifdef for MacOS X support. Yes, MacOS :-)
	  The "Developers CD"/Kit is required to compile.

	Work around BUG in XFree Xaw libraries: They dont accept
	 widechar cut-n-paste. So access clipboard 'directly'.
	[cut-n-paste was working fine under Solaris all this time]
	Good news is, code is somewhat cleaner now anyway.

5.12.2
	Minor bug in DisplayLearnChar() fixed.
	Fixed Handle_paste() to handle paste from kterm, as well as
	just netscape.


6.0
	Yes, that's right... the legendary radical search has arrived!!


6.0.1	Made loading of radkfile more robust. Since I "discovered"
	that feeding it a bad file, made things really difficult
	to trace down.

6.0.2   Added -radkfile option, just to be consistent.
	Added jya,jyu,jyo as synonyms for ja,ju,jo in romaji input.
	Also added tzu to shadow du.
	Fixed bad "zu" mapping in "display romaji" mode, for du.

6.1	Added special middle-button click on the search results,
	that splits up a multi-kanji string into individual kanji in the
	multi display popup.

6.1.1	Fixed major bug in InitRadicals: radical table was not getting
	read into memory correctly.

6.1.2	Minor tweaks to search window layout

6.2	Made radicalsearch window disable impossible combinations
	Fixed major bug in radical searching, that "lost" radical matches
	Majorly pinned down widget locations in almost all windows
	 It seems that recent versions of xaw, on multiple platforms,
	 have broken proper XawRubber constraints for forms.
	 Resizing will eventually migrate all widgets offscreen to
	 the top left. Sigh.
	[internal code] Tried to make brace indentation more consistent across
	  the files.  Went through every durn file. phew.

6.2.1	Bumped radical search max display limit cutoff to 200

6.2.2	Tweak to resize on main window and search window.
	Resizing is now a lot more fixed, to avoid distortion.
	See 6.2 comments, above

	Minor tweak to UpdateSearchlabels, to initialize when no radfile.
	Fixed font use in 'skip' window

	Attempted to set "busy" cursor in SelectRadical().
	Except that Xaw wont allow that sort of thing while in a
	  callback routine, I guess. Need a WorkProc instead.

6.3	added handling of non-shifted JIS for cut-n-paste (rxvt support)

6.3.1	Added handling of ascii-escaped unicode char cut-n-paste
	 (for pan, and other gtk2 based progs)
	Added "KDrill*international: False" resource entry (again?)

	Made multikanji reset scroll position on new search
	Tweaked multikanji usefile toggle some.
	Changed search method popups to be "topLevel" class, so it is
	 possible to iconify main kdrill window and still search fully.

6.4	Changed the "missed" window, to display
	(unique wrong)/(wrong click count)/(guesses)
	Plus fixed nasty bug where "dismissed" popups dont come back,
	probably introduced as result of the popup changes in 6.3.1

6.5	patched array size on SKIPshapeinput to be [4] instead of [3].
	Thanks David Binderman for noticing and reporting this.
	Fixed DoStrokeSearch to have missing NULL for XtVaSetvalues(),
	Thanks to Mike FABIAN for pointing this out.
	Patched doJISConvert to understand gtk2 unicode cutnpaste format.
	Patched romaji input tables to accept xtsu for small-tsu
	Patched code so that the showinorder toggle gets saved to prefs file