On Thu, Nov 11, 2010 at 22:50, Rimas Kudelis <[email protected]> wrote: > 2010.11.11 16:08, Aron Xu rašė: >> >> On Thu, Nov 11, 2010 at 20:50, Rimas Kudelis<[email protected]> wrote: >>> >>> 2010.11.11 13:21, Aron Xu rašė: >>>> >>>> Yes there is already a zh-hans item on that page. I'm not sure whether >>>> we'd change it to zh_CN, because in glibc our language code is zh_CN, >>>> and Mozilla use zh-CN. "zh-hans" is the non-official form to express >>>> the combination of zh_CN and zh_SC, similarly "zh-trans" is for zh_TW >>>> and zh_HK. >>> >>> Dunno where you took that info from. zh-Hans (means Chinese in Han script >>> Simplified variant) is an official code from BCP47, which should be >>> preferred to zh-CN (Chinese in China). >>> >>> Similarly, zh-Hant (Chinese in Han script Traditional variant) are >>> preferred >>> to zh-TW and zh-HK. >>> >>> Not all environments are already using these new codes, but in general, >>> the >>> direction of movement is towards them, not from them. For example, Apple >>> and >>> Microsoft have introduced them in their products recently. >>> >>> Regardless of said above, I'm quite positive that we should take OS >>> expectations into account, and if zh-Han* locale codes aren't recognised >>> by >>> Linux, our Linux packages should probably use older, recognised locale >>> codes. >>> >>> Rimas >>> >> I'm not so familiar about BCP 47 but I heard it was designed for >> Internet application usage (such as HTML). So I agree to use it in our >> wiki and other web documentations. > > Well, I haven't read the whole standard either, but I don't think Internet > is its only application. Similarly, MIME also contains a letter for > Internet, but it's being used way more widely. ;) > >> But for the software, I think we are using the ISO 639-1/2 in most >> cases. Lists of languages in these two ISO standards are [1] and [2]. > > ISO 639 defines languages, not locales or scripts. Basically, BCP 47 > combines its codes from those defined in ISO 639, ISO 15924, and ISO 3166. >
Ah, yes. >> I think LO is a general desktop application suit and should follow a >> standard that is well accepted, there won't be a better choice than >> using ISO 639-1/2, which is compatible (or almost compatible) to most >> platforms (Windows[3], Mac[4], Linux[5] and other *nix variants). > > I don't see zh-CN or zh-TW defined in ISO 639. ;) Code zh is defined, but > the second part – CN and TW – comes from a totally different standard (ISO > 3166). > Similarly, Hans and Hant are defined in ISO 15924. > > Also, check out [6] for Mac. > But keep in mind Hans and Hant cannot cover different kinds of Chinese, for example there are differences between HK and TW, but they are combined to one single Hant, which perhaps could not be accepted by people who are using them, :) >> BCP 47 is used in OOo (correct me if I'm wrong!), but I don't think >> it's a wise choice because we have to first map ISO 639 codes to >> BCP-47 for we use gettext have i18n support, then we have to map BCP >> 47 back to Unix locales (which is almost ISO 639 codes) on most *nix >> platforms. Such a process is complicated. > > I don't think we're using gettext (yet). > I'm not sure about this point, things should be not so difficult if we do not use it. >> Language code usage of software are in a mess. Chinese on Windows, for >> example, we can find: >> * zh_CN (ISO 639-1) for Windows itself (as in [3]); > > Well, [3] lists codes for Chinese in different territories, not for chinese > written in different scripts. > >> * zh-Hans (BCP 47) in Vista and .NET 2.0; >> * zh-CHS which should be replaced by zh-Hans but still being widely >> used till even today because of Windows XP; > > From what I've read, it's going to be obsoleted in future versions of .Net. > Yes, I've wrote that it is to be replaced, but remember Windows XP is still running on many people's desktops and laptops, I'm not sure this situation can change very rapidly in near future. And some other applications are still using zh-CHS as a tag on there release documents to tell people it is in Simplified Chinese, no matter whether they have to use zh-Hans in new .NET, they just show such information to end users and people get confused. >> * zho (ISO 639-2) are being used in some Microsoft documents as well. >> >> There is no doubt that we should follow ISO 639-1 because all >> standards mentioned above are based on it. For languages that do not >> have an ISO 639-1 code, I suggest we use ISO 639-2 so we can easily >> use gettext and support *nix systems, and map the codes to respective >> platforms (Windows, Mac, etc) if necessary. > > That's exactly what BCP 47 does. The "zh" part of zh-Hans is exactly the > code assigned for Chinese by ISO 639-1. ;) > Well, I don't mind BCP 47 if it works well, but I've mentioned before that it cannot cover all different variants, it's just a loosely defined standard. To be more precise to end users, we might have to place two sets of translations, zh_TW and zh_HK, into zh-Hant packages. They are not the same, so we need to do it separately, even if we make them into a single package to end users. Listing language variants with different regions is a good way to solve conflicts in our development. On the other hand, using a loosely defined name for our release language pack (which contains everything fit into the category) is probably good for users. >> [1]http://en.wikipedia.org/wiki/List_of_ISO_639-1_codes >> [2]http://en.wikipedia.org/wiki/List_of_ISO_639-2_codes >> [3]http://msdn.microsoft.com/en-us/library/ms533052 >> [4]http://support.apple.com/kb/TA26811?viewlocale=en_US >> [5]http://sourceware.org/git/?p=glibc.git;a=tree;f=localedata/locales > > [6] > http://developer.apple.com/library/mac/#documentation/MacOSX/Conceptual/BPInternational/Articles/LanguageDesignations.html > > > Rimas > > > -- > E-mail to [email protected] for instructions on how to unsubscribe > List archives are available at http://www.libreoffice.org/lists/l10n/ > All messages you send to this list will be publicly archived and cannot be > deleted > > -- Regards, Aron Xu -- E-mail to [email protected] for instructions on how to unsubscribe List archives are available at http://www.libreoffice.org/lists/l10n/ All messages you send to this list will be publicly archived and cannot be deleted
