Journey To a Localised Free Desktop

Danilo Šegan

danilo@gnome.org

http://danilo.segan.org/

What Is Localisation?

• adjusting the system to suit (y)our locale
• what the hell is "locale"?
• why do we want it?
  • broader user base
  • more accessible
  • emotional and cultural
• how do we achieve it: that's what we're here to discuss, so please fasten your seat belts!

Relevant standards

• Codes: ISO639, ISO3166
• Character encoding: ISO10646, Unicode
• System support: POSIX, ISO14561, ISO14562
• Locale databases: CLDR, GNU libc
• (Too) many others...

Locales

• Introduction: their purpose, data, CLDR
• Locale naming in GNU libc: ll_CC@modifier:

  ll

  language code, most commonly ISO 639-1 (alpha-2), but sometimes ISO 639-2 code (alpha-3)

  CC

  ISO 3166-1 country code

  modifier

  free form modifier, might indicate currency (euro), script (Latn or latin) or anything else

Simple Locale Definition

• Using "copy" keyword
• LC_COLLATE "iso14651_t1" data set

Example

comment_char %
escape_char  /

LC_TIME
day     "<U0044><U0061><U0079><U0020><U0031>";/
        "<U0044><U0061><U0079><U0020><U0032>";/
        "<U0044><U0061><U0079><U0020><U0033>";/
        "<U0044><U0061><U0079><U0020><U0034>";/
        "<U0044><U0061><U0079><U0020><U0035>";/
        "<U0044><U0061><U0079><U0020><U0036>";/
        "<U0044><U0061><U0079><U0020><U0037>"
END LC_TIME

LC_COLLATE
copy "iso14651_t1"
END LC_COLLATE

% Repeat for CTYPE, MESSAGES, MONETARY, NUMERIC, PAPER, 
% 	     MEASUREMENT, NAME, TELEPHONE, ADDRESS
LC_IDENTIFICATION
copy "de_DE"
END LC_IDENTIFICATION

...

Managing and Using Locales

• localedef command:

  localedef -c -f UTF-8 -i test test.UTF-8

• LANG and LANGUAGE variables
• LC_* variables (LC_MESSAGES, LC_TIME, LC_COLLATE, LC_MONETARY, LC_NUMERIC

Displaying localised content

• Fonts
• Rendering Technologies
• Display Problems (word- and line-breaking, positioning)

Fonts and their development

• FontForge and Mensis: http://fontforge.sf.net/
• OpenType features
• Choosing the Right Fonts
• FontConfig and ~/.fonts/

Drawing text

• FreeType2: font file handling
• Pango: glyph positioning and combining
• Gtk+: making it all work

Right-to-Left Display

• Common issues: UI elements, mixed content
• Gtk+ Tip: translation of "default: LTR" message

Localised Keyboard Input

• RTL
• Complex (non-alphabetic) Scripts
• Keyboard Shortcuts
• Interesting Technologies: Linux console, GNU Hurd console, xmodmap, XKB, IMs, Compose mechanisms

XKB

• Terminology and Concepts: layouts, groups, shift levels, modifiers
• Short Sample

  default partial alphanumeric_keys
  xkb_symbols "basic" {
    name[Group1]= "Test keyboard layout";
  
    include "pc/latin(basic)"
    include "pc/test(atoz)"
    include "pc/test(level3)"
  };
  
  partial alphanumeric_keys
  xkb_symbols "atoz" {
    key.type[Group1] = "FOUR_LEVEL_ALPHABETIC";
    key <AC01> {   [ z,        Z,   any,any ]   };
  };
  
  partial alphanumeric_keys
  xkb_symbols "level3" {
    key <AC01> {   [ any,any,     dead_acute,          U301 ]   };
  };

• Further help: xkeyboard-config and i18n@xfree86.org lists

Input Methods

• Gtk+ input modules
• XIM
• SCIM
• Specialised IMs: XCIN for Chinese

Translation

• Basic Concepts
• Common Problems:
  • context: directions for translators
  • message contraction
  • mark-up
  • plural forms
  • numbers and monetary values
• Minor differences

Gettext and PO files

• Gettext Basics: translation domains, PO, MO files
• PO file features:
  • English as a key for lookup
  • Context for translators (source references, comments,...)
  • "Tags" (fuzzy matching, syntax checking,...)
• intltool: single PO for everything (and context markers)
• multi- and external-domain translation

Translation in Gtk+, Glade, Gnome

• Gtk+ caveats:
  • setting locales
  • activating translations (GtkUIManager and GtkActionGroup: set_translate_func, set_translation_domain), GtkItemFactory: deprecated, similar)
  • Glade and libglade tips: might need to specify translation domain in call to glade_xml_new()

Documentation Translation

• xml2po and further goals (intltool integration, diff syntax):

  #! Added [-plugin -]%s{+ here+} with %d %s.[-\n-]
  #: tools/gst-register.c:49
  #, c-format, fuzzy
  msgid "Added %s here with %d %s."
  msgstr "Придодат додатак %s са %d %s.\n"
  

• gnome-doc-utils localised stylesheets
• yelp

Other Issues

• Spell-checking: Aspell as a solution:
  • works for any suitable language (as of 0.60)
  • word lists
  • extended features: http://aspell.net
• Accessibility and usability:
  • speech recognition
  • speech synthesis
  • on-screen keyboards (hint, hint: GOK)

Conclusion

• Danke
• Localising in your $HOME
• Free-form session on docs tommorow with the all-familiar Shaun McCance (and we can discuss documentation translation as well)


• No autographs today, sorry