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

Re: Help about localization packages



Quoting Vincent Danjean (vdanjean.ml@free.fr):

>   Upstream store its language dependent files in repository with
> non-classical names. For example: ar-Arab, bg-Cyrl, fr-FR, zh-Hans-CN, ...
> I made a debian/rules target to automatically update debian/control
> and debian/koha-l10n-XXX.install files (this target is not used for
> standard builds: it must be triggered manually by the maintainer).
>   But, to do that, I need a relation between
> - the upstream name (ie zh-Hans-CN)
> - the package name [after koha-l10n-] (ie zh-cn)
> - the language name [for the Description] (ie "Chinese (China)")

I am very strongly opposed to localization wearing the country along
with language name. I'm perfectly aware that the way languages are
spoken varies from one place to another but the only strong enough
exception I've seen for this is "pt" ("general" Portuguese)
vs. "pt_BR" (Brazilian Portuguese).

All others do not really need to add a country code after the language
code, so I'll comment with that in mind.

So, the koha l10n packages should (if not must) use "koha-l10n-xx"
with "xx" being the ISO-639 code of the language.

The only three possible exceptions I would see are:

koha-l10n-pt-BR
koha-l10n-zh-cn
koha-l10n-zh-tw

Of course the very quirky way of upstream to name its files does not
help. Have they ever heard about ISO 639 and ISO 3166?

> 
> To do that, I begin with the help of the iceweasel-l10n-* packages.

Please note that some of them are, imho, incorrect:

iceweasel-l10n-uk-ua
iceweasel-l10n-cy-gb
iceweasel-l10n-dz-bt
iceweasel-l10n-et-ee
.../...
indeed all those using a country code modifier, except pt-br, zh-cn
and zh-tw


> I applied the following rules :
> A) if I can identify the language from upstream name and iceweasel packages,
>    I use the same package name and language name.
>    => it means that most of package names are the first two letter of
>       upstream name but it is not an absolute rule (even when there seems
>       to have no ambiguity such as with 'gl' -> 'gl-es',
>       'fa-Arab' -> 'fa-ir')
>    Can you check if my guesses are correct (and, if it exists, tell me
>    where I can find a list with all 'extension') ?
> 
> B) there are some upstream name for which I do not know the corresponding
>    debian name and language name.
>    Can you help me for these ?
> 
> The debian/* files (control and rules in particular) can be seen here:
> http://git.debian.org/?p=collab-maint/koha.git;a=tree;h=refs/heads/wip;f=debian
> 
> Relevant information extracted from debian/rules:
> # upstream name I do not know how to interpret them (case B):
> UNSUPPORTED_LANGS=am-Ethi es-VE hy-Armn kn-Knda lo-Laoo mi-NZ tet ur-Arab

am=Amharic: localization package should be koha-l10n-am
es-VE stands for "Spanish in Venezuela". I would personnally advise
against using that and just use "koha-l10n-es" by using what they call
es_ES upstream
hy-Armn: Armenian. Localization package should be koha-l10n-hy
kn-Knda: Kannada (language from India, state of Karnataka mostly)
         --> koha-l10n-kn
lo-Laoo: Lao --> koha-l10n-lo
mi-NZ: Maori --> koha-l10n-mi
tet: Tetun, a language from Indonesia. That language has fairly few
     speakers. Is that what upstream intended?
ur-Arab: Urdu. Nothing to do with "Arab". Urdu is a national
         language of Pakistant --> koha-l10n-ur
    



> 
> # LL_NAME_(upstream name) is my guessed language name (for description)
> # DEB_NAME_(upstream name) is my guessed debian name (when this is not
> #   the first two letter of upstream name)
> LL_NAME_ar-Arab=Arabic

koha-l10n-ar

All your other guesses are correct.


Attachment: signature.asc
Description: Digital signature


Reply to: