proc slide args {
   global slides
   if {![info exist slides]} slide'init
   incr slides(N)
   set slides(title,$slides(N)) [join $args]
 }
 proc slide'line {type args} {
   global slides
   lappend slides(body,$slides(N)) [list $type [join $args]]
 }
 foreach name {* + -} {interp alias {} $name {} slide'line $name}

 proc slide'init {} {
   global slides
   array set slides {
      canvas .c  N 0  show 1 dy 20
      titlefont {Tahoma 14 bold} * {Tahoma 9 bold} + {Tahoma 8}
      - {Courier 9}
   }
   pack [canvas .c -bg white]
   foreach e {<1> <Right>} {bind . $e {slide'show 1}}
   foreach e {<3> <Left>} {bind . $e {slide'show -1}}
   wm geometry . +0+0
   after idle {slide'show 0}
 }
 proc slide'show delta {
   upvar #0 slides s
   incr s(show) $delta
   if {$s(show)<1 || $s(show)>$s(N)} {
      incr s(show) [expr -$delta]
   } else {
      set c $s(canvas)
      $c delete all
      set x 10; set y 20
      $c create text $x $y -anchor w -text $s(title,$s(show))\
         -font $s(titlefont) -fill blue
      incr y $s(dy)
      $c create line $x $y 2048 $y -fill red -width 4
      foreach line $s(body,$s(show)) {
         foreach {type text} $line break
         incr y $s(dy)
         $c create text $x $y -anchor w -text $text \
         -font $s($type)
      }
   }
 }
 bind . <Up> {exec wish $argv0 &; exit} ;# dev helper
#--------------- The rest is data - or is it code? Anyway, here's my show:
 slide i18n - Tcl for the world
 + Richard Suchenwirth, Nuremberg 2003
 +
 * i18n: internationalization
 + 'make software work with many languages'
 +
 * l10n: localization
 + 'make software work with the local language'

 slide Terminology
 * Glyphs: 
 + visible elements of writing
 * Characters: 
 + abstract elements of writing
 * Byte sequences: 
 + physical text data representation
 * Rendering: character -> glyph
 * Encoding: character <-> byte sequence

 slide Before Unicode
 * Bacon (1580), Baudot: 5-bit encodings
 * Fieldata (6 bits), EBCDIC (8 bits)
 * ASCII (7 bits)
 + world-wide "kernel" of encodings
 * 8-bit codepages: DOS, Mac, Windows
 * ISO 8859-x: 16 varieties

 slide East Asia
 * Thousands of characters/country
 + Solution: use 2 bytes, 94x94 matrix
 + Japan: JIS C-6226/0208/0212
 + China: GB2312-80
 + Korea: KS-C 5601
 +
 * coexist with ASCII in EUC encodings

 slide Unicode covers all
 * Common standard of software industry
 * kept in synch with ISO 10646
 + Used to be 16 bits, until U 3.1
 + Now needs up to 31 bits
 * Byte order problem:
 + little-endian/big-endian
 + U+FEFF "Byte Order Mark"
 + U+FFFE --illegal--

 slide UTF-8
 * Varying length: 1..3(..6) bytes
 + 1 byte: ASCII
 + 2 bytes: pages 00..07, Alphabets
 + 3 bytes: pages 08..FF, rest of BMP
 + >3 bytes: higher pages
 +
 * Standard in XML, coming in Unix

 slide Tcl i18n
 * Everything is a Unicode string (BMP)
 + internal rep: UTF-8/UCS-2
 * Important commands:
 - fconfigure \$ch -encoding \$e
 - encoding convertfrom \$e \$s
 - encoding convertto   \$e \$s
 +
 * msgcat supports l10n: 
 - {"File" -> [mc "File"]}

 slide Tk i18n
 * Any widget text is Unicoded
 * Automatic font finding
 + Fonts must be provided by system
 +
 * Missing: bidi treatment
 + right-to-left conversion (ar,he)

 slide Input i18n
 * Keyboard rebinding (bindtags)
 * East Asia: keyboard buffering
 + Menu selection for ambiguities
 +
 * Virtual keyboard (buttons, canvas)
 * String conversion: *lish family
 - {[ruslish Moskva]-[greeklish Aqh'nai]}

 slide i18n - Tcl for the world
 +
 +
 + Thank you.