On 17/07/2008, at 9:02 PM, Christian Perrier wrote:
Some translators may remember the need to fill in a file named after their language code in the "installer/build/needed-characters" directory of D-I SVN.
You can check that quickly by following this link: http://svn.debian.org/wsvn/d-i/trunk/installer/build/needed-characters/
Up to now, the policy for these files was to ask you the characters that are *needed* to properly write your language and that are not part of the ASCII range. For instance, for French: àâéèëêïîôùûüÿÀÂÉÈËÊÔÙÛÜŸÇ° Polish: ąĄćĆęĘłŁńŃóÓśŚżŻźŹ etc.After some discussion, we just decided to extend this a little to *all characters* that can be typed on your respective keyboards. This way, even when users input strings with such characters, something will be displayed (otherwise, there was a chance that, the character being stripped out from the console font, nothing coul dbe displayed.So, please take care to update the file relevant to your language in installer/build/needed-characters, with characters that can be typed on your keymap *and that are printed on the keys* (no characters that are "hidden" on some key combinations: the point is allowing all characters we *reasonably* expect users to enter).Frans Pop and I will monitor such changes and will nag you if we think that you're adding an unacceptable number of characters, so please be conservative enough.
My language has 72 vowels, 66 of them input with dead-key combinations (varying with the layout and input system) and not represented on any physical keys.
I notice (in the file "vi.utf") that they're all already listed, but it's a good example of a language having a large number of characters in this situation.
from Clytie Vietnamese Free Software Translation Team http://vnoss.net/dokuwiki/doku.php?id=projects:l10n
PGP.sig
Description: This is a digitally signed message part
