[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-i18n-doc
Subject:    Re: KDE 3 language codes
From:       Gaute Hvoslef Kvalnes <ai98ghk () stud ! hib ! no>
Date:       2001-12-03 13:57:26
[Download RAW message or body]

-----Original Message-----
From: Eric Bischoff
To: Frederik Fouvry; kde-i18n-doc@kde.org
Sent: 03.12.01 13:39
Subject: Re: KDE 3 language codes

> Currently, we do not conform neither to the RFC nor to ISO-639-2,

We can't conform to ISO 639, since it's just a list. The choice is 
between conforming to RFC 3066 and ISO/IEC 15897, and conforming to 
a custom KDE standard.

> I find the idea of
> 
> 	"ISO-639-2 gives the codes but doesn't say you must use them,
> 	the RFC says that the codes exist but that no one should use
> them"
> 
> rather schyzophrenic.

It's not, the ISO codes just has a broader scope than this particular 
RFC. RFC 3066 and ISO/IEC 15897 defines the usage of language codes 
for purposes like ours. If we decide to use only ISO 639-2 codes, 
then we have 'invented' our own standard.

I don't see why KDE should do things its own way when there already 
exists a perfectly good standard.

- We have to create a mapping to official standards anyway, for 
HTTP negotiation (and XML docs?).

- We just have to change a few codes if we stick to RFC 3066, while 
Eric's suggestion requires changing almost all codes.

- We have to create a backwards mapping as well, to KDE's old 
codes, for old KDE applications and third-party applications.

- We *should* pay attention to the language settings in Linux. For 
example, KPersonalizer should suggest the default Linux language as 
default KDE language when starting KDE for the first time. This is 
much simpler if we use the same codes as everybody else.

> Frederik also says that the RFC provides no hardwired mechanism for 
> distinguishing "pt_BR" from "pt" or "zh_CN" from "zh_TW", and that this 
> distinction should be based on a fallback mechanism. I think this is
> another 
> excellent reason for not sticking to this braindead RFC ;-). "zh_CN" is
> _not_ 
> a fallback of "zh_TW", nor the contrary, and I don't want to start a war
> in the Formose detroit for having assumed that ;-).

The codes themselves should not suggest fallback languages at all. The 
locale settings should take care of that, and in some cases KDE can 
provide reasonable default fallbacks. In Norway, Bokmål and Nynorsk 
should be fallbacks for eachother; while in Brazil, perhaps 'ordinary' 
Portuguese is a sensible fallback for Brazilian Portuguese? More 
importantly, this has to be user definable.

Regards,
 Gaute Hvoslef Kvalnes

[prev in list] [next in list] [prev in thread] [next in thread]