[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Conference at debconf on i18n/l10n: ideas? suggestions?



Hi,

> I'm not sure if there is anyone. In any case those issues are not 
> completely debian-specific. However I will make sure to point to your 
> excellent DDP document (intro-i18n) for encoding issues (which I don't 
> master, BTW)

Thank you for considering my opinion.

I don't require Debian developers to master i18n programming.  I think
it is apparently too heavy.  However, I want Debian developers that
there exist such problems, so that when they received bug reports or
patches related to i18n, they understand what the reporters are saying
and how severe the problem is.  It is like that security is not Debian-
specific problem but every Debian developers should know at least basic
concept of security, though I agree that I18N is less important than
security.

I imagine the last screenshot of my debconf webpage or screenshots from
my "mojibake" webpage:
http://www.debian.or.jp/~kubota/mojibake/
will help developers feel the severity of the problem and the situation
where Asian people are located.  I hope people who see the screenshots
will understand why CJK people says "translation is a relatively small
topic in i18n area".

Thus, please don't hesitate even if you don't master encoding issues.
If there will be some technical details, I think the point is that
developers should not depend on assumptions which are valid only for
a part of languages (such as "one character is one byte", "words are
separated by whitespaces", and so on).

For example, it is apparently inappropriate to use strlen() to count
number of characters in a string.  Instead, there are several ways:
  (1) convert any strings to/from wchar_t on I/O and use wchar_t
      for all internal processing (please see textbooks on ISO C
      for detail),
  (2) convert any strings to/from Unicode on I/O and use Unicode
      for all internal processing (please read documents on Unicode
      and iconv(3) for detail), or
  (3) hard-code all encodigs on this planet (not recommended).

BTW, the descriptions in intro-i18n get older and older these days.
I have to revise....

---
Tomohiro KUBOTA <kubota@debian.org>
http://www.debian.or.jp/~kubota/




Reply to: