R help - Feb 2005 - Internationalization and localization of R

This pre-announcement is being sent for information to both R-help and 
BioC.  Please use R-devel at r-project.org for any follow-up discussion.

-----------------------------------------------------------------------

We are about at the end of a several-week process of adding support for
non-Latin character sets and non-Western-European languages to R.

- For Linux users, R works in the UTF-8 locales that are rapidly becoming
   the standard in the latest distributions: for example if in Fedora Core
   3 I select English (United Kingdom) I get locale en_GB.utf8.  This is
   also true of those commercial Unixes which support such locales.

- For Linux/Unix users, R works in many non-English locales with other
   encodings such as EUC-JP.  The only one that we are having problems with
   is vi_VN.tcvn (but the Vietnamese UTF-8 locale appears to work).

- For Windows users, R works in the double-byte `East Asian' locales as
   well as many others, if support is selected at installation time.
   (See the current rw-FAQ for more discussion of this: this is supported
   in the current automated builds of r-devel on CRAN.)

- There are built-in facilities to convert between encodings, so that e.g.
   Japanese in SHIFT-JIS can be read by R running under UTF-8 or EUC-JP.

- There are mechanisms to translate both C and R error messages from both
   base R and for packages, and also the menus etc of the Windows and
   MacOS X GUIs.

The purpose of this pre-announcement is to alert developers that these 
changes will be coming in 2.1.0 in a couple of months.

- Package writers who would like to translate messages in their packages,
   or to prepare their packages for translation by others should consult

   http://developer.r-project.org/210update.txt

   and the latest version of `Writing R Extensions' in the R-devel sources.

   All the standard packages and a few of the recommended ones have been
   prepared and updates will be appearing on CRAN shortly.

- People might like to start organizing translation teams for their own
   languages, and early experience from such a team would be helpful in
   polishing the instructions and mechanisms.  There is a document for
   translators at

   http://developer.r-project.org/Translations.html

   If translations are available by early April they can be shipped with
   2.1.0 (although there are planned to be mechanisms to add them to an R
   installation).

- We will be looking for testers in the alpha/beta period in about a
   month's time, and would-be users of R in unusual languages (e.g.
   Vietnamese) would be especially helpful as testers.  This is
   particularly important for Windows 95/98/ME to which we only have
   very limited access, to English versions.

-----------------------------------------------------------------------

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

Prof Brian Ripley writes:
> - People might like to start organizing translation teams for their own
>   languages, and early experience from such a team would be helpful in
>   polishing the instructions and mechanisms.  There is a document for
>   translators at
> 
>   http://developer.r-project.org/Translations.html
> 
>   If translations are available by early April they can be shipped with
>   2.1.0 (although there are planned to be mechanisms to add them to an R
>   installation).
> 
        I?ve been playing with the translation files here (I intend to
to translate the base and recommended packages files to Portuguese), and
I noticed a minor problem with the instructions detailed in 

        http://developer.r-project.org/Translations.html

        Where it says:

" Then compile and install the translated catalogues by 

mkdir ../inst/po/LC_MESSAGES/sl
msgfmt -o ../inst/po/LC_MESSAGES/sl/R-splines.mo R-sl.po
msgfmt -o ../inst/po/LC_MESSAGES/sl/splines.mo sl.po
"

        It should read:

"
(...)
mkdir ../inst/po/sl/LC_MESSAGES/
msgfmt -o ../inst/po/sl/LC_MESSAGES/R-splines.mo R-sl.po
msgfmt -o ../inst/po/sl/LC_MESSAGES/splines.mo sl.po
"

        After fixing this I managed to get the gettext translated
messages to work here. 

        On a side note, are there any plans regarding the translation
of the .Rd documentation files? As much as the translation of
messages is important, I believe that the online documentation would
also be an essencial issue.



--
Fernando Henrique Ferraz P. da Rosa
http://www.ime.usp.br/~feferraz