On Sunday 30 March 2003 09:48 am, Keld J�rn Simonsen wrote: > On Sat, Mar 29, 2003 at 08:33:02PM -0800, Edward Cherlin wrote: > > On Friday 28 March 2003 01:21 pm, Keld J�rn Simonsen wrote: > > > I agree with Kubota-san and Peter, Internationalization > > > should be inherent in all programs, and even American > > > programmers should be able to easily write > > > internationalized programs. > > > > > > One idea I have had was that strings in programming > > > languages should automatically be put for translation, > > > unless it is a constant. > > > > > > Is that a scheme that would work? > > > > Not in that simplistic form. Programmers frequently compose > > messages from pieces that fit together in the language and > > context they are most familiar with, but not in others. > > Variations in the way languages deal with gender, number, > > declensions, conjugations, sentence order, polite speech, > > and other factors interfere. > > Of cause there needs to be some explicit i18n functionality > adressing this, but that is probably a level that we cannot > make i18n-na�ve programmers learn easily. > > My idea was that strings in programs automatically were > internationalized and there were readily available tools like > .po-files available for localization. That would satisfy some > part of Kubota-sans goal. What would be lost in this?
It is a very positive step forward. The question was whether it would be enough, and we agree that it would not. > This would also lead to cleaner programming, removing some of > the i18n overhead in the syntax, like explicit calls of the > gettext macro. > > One could then have a macro to mark a text *not* for > translation. Essential for language tutorial and reference works, certainly. > > The classic Chomskian example of sentences with the same > > word order but different structure is the two sentences > > Time flies like an arrow. > > Fruit flies like a banana. > > > > In English one can say, "Is it here?" or "Is he here?" > > However, in Japanese those ideas are expressed with distinct > > verbs, "aru" for the inanimate and "iru" for the animate. > > > > My favorite example is the Japanese utterance, "Boku wa, ebi > > da." After the first few months of studying Japanese, an > > American would attempt to translate this as, "I am a > > shrimp." Actually, when spoken to a waiter, it means, "Mine > > is the shrimp dish." > > Yes, there are examples like this, but they are not frequent > in Open Source translations. I have personally been thru > hundred of thousands of translations (for Danish) and from my > memory the problem of ambiguity occurs in less than 1 % of the > strings for that language. I believe that one of the main reasons why you do not encounter this problem is that the Linux Localization community has been over this ground many times, so that old broken code has been fixed, and Linux developers have been tolerably well educated. Nevertheless, the problem reappears with great frequency when internationalizing code from more backward platforms. > But of cause there should be tools to address the issue, and > in my mind that should be built into the programming language. > > My idea was to address this first for the programming > languages C and C++. Comments? I would be interested to see your proposal, and to review it with an eye to improvement. > > > Could we just do some automated tools to mark every string > > > for translation via gettext - or would it need further > > > spec, like getting it thru some standards process? > > > > No, we actually need to teach them how to do it right. The > > details are application-specific. > > Yes, we could also provide tutorials, but I fear that many > programmers would not take the trouble to read it and follow > them. On the other hand almost all applications I use today > under Linux are internationalizaed and loclalized for my > language. > > Are there any good tutorials out there for how to do i18n > programmimng? Nadine Kano wrote one, published by Microsoft, which is unfortunately very much out of date and out of print. I know of one for Visual Basic, and a few magazine articles. There is a lot of material from commercial software vendors about the use of their proprietary tools, such as translation memory systems. Perhaps some of us should get together and pitch the idea to O'Reilly. Certainly a HOWTO is in order. > Best regards > keld -- Edward Cherlin Generalist & activist--Linux, languages, literacy and more "A knot! Oh, do let me help to undo it!" --Alice in Wonderland -- Linux-UTF8: i18n of Linux on all levels Archive: http://mail.nl.linux.org/linux-utf8/
