prev Mailing List xkb@listserv.bat.ru Message #104 next

From: Danilo Segan <dsegan@gmx.net>
Subject: Re: is gettext 0.14 supported by intltoolize
Date: Thu, 01 Jul 2004 11:03:14 +0200
To: "Sergey V. Udaltsov" <svu@gnome.org>
Cc: xkb@listserv.bat.ru
Full Headers
Undecoded Letter
Hi Sergey,

Данас у 0:56, Sergey V. Udaltsov написа:

>> Meanwhile, I've noticed several problems in Serbian layouts I created
>> (and which are included in xkeyboard-config). First off, "sr" was
>> renamed to "scc" (three-letter ISO639 code for language, though
>> Serbian is among few languages which have another three-letter code
>> "srp" in ISO639), which sucks, since nobody is going to look under
>> "scc" for Serbian ("srp" is much better alternative, and it is also
>> ISO639!).
> OK. Agreed. Fixing it right away. It will be "srp".

Thanks. I see you've already fixed all the references in "yu" (now
"cs") as well. Thanks for doing that too.

>> It also broke "yu" layout, which depends on "sr" one. Besides,
>> there's no such country as "yu" — it should now be "cs" (if, otoh,
>> you're going to leave "yu" as is, then I suggest you also name Serbian
>> layout "sr" — Suriname users, in case they show up can use any of the
>> three-letter choices offered at
>> http://www.ethnologue.com/show_country.asp?name=Suriname appropriate
>> for their respective language).
> Well. There is some mistery here. "cs" is for "Czech" in iso639, isn't
> it? What does it have to do with "yu"??? Also, what could be
> "shortDescription" and "description" tag values for "cs"?

"CS" is ISO3166 code for Serbia and Montenegro since July 2003 (I
guess the reasoning was that Serbia is "Srbija", and Montenegro is
"Crna Gora", so you get "Crnagora-Srbija" ;). New alpha-3 ISO3166
code for Serbia and montenegro is "SCG".

"YU", otoh, is old, pre-July 2003 code for a country of Yugoslavia,
which changed name in February 2003 to Serbia and Montenegro.

See following link for official statement from ISO:
http://www.iso.ch/iso/en/prods-services/iso3166ma/01whats-new/2003-07-23_statement_cs.html

I also based my reasoning on the fact that you seem to be using
alpha-2 (two-letter codes) in case of countries, and alpha-3 in case
of languages (so "cs" from ISO639 doesn't count, since ISO3166 code
for Czech is "cz" ;). So, correct country code to use for Serbia and
Montenegro is "CS", and correct language code is either "srp" or "scc"
(though first is preferred, because it's much easier to recognize).

There're also some subtle problems here. As you said on the web page
along with the release of 0.1:
    Each language/country has no more than one layout and any number
    of variants.

I want to help with that goal as well, but the problem here is that
Serbian in Serbia and Montenegro is officially written using Cyrillic
script. Latin script has no mention except that it is used by
minorities. Thus, there would be no place for it in either "cs" or
"srp". Yet, users are really expecting Latin variant to be available
for them to use, so we cannot just drop it.

If we're to have a single file with all the variants, it should
probably be "cs", but it's going to cause much confusion — Czechs
might think that it's for their native language, and Serbs won't
know where to look to find their keymap (well, of course, they could
use a tool such as gswitchit, but many probably won't, and will have
hard time finding out what map they want).

I just joined xkb@listserv.bat.ru, so perhaps we should continue this
discussion over there? (I'm CCing the list, I hope you don't mind me
doing that.) I'm expecting Frank Murphy might have some comments,
since we discussed similar issues at great length on i18n@xfree86.org
list in the past.

For that sake, I'd repeat my proposal I gave at that time: use
case-sensitive filenames, with eg. uppercase being country code, and
lowercase being language code. That would allow for things like "cs"
for Czech, and "CS" or "SCG" for Serbia and Montenegro (no need to
insist on only one being allowed — case is what helps us distinguish
them), without allowing for confusion (since we'd choose "SCG" in this
case, or simply "sr"). Since X works only on case-sensitive systems
that I know of, I don't see a problem with this (even GNU libc uses
uppercase for language codes in locale names).

> Also, I'll play with po/Makefile.am to add update-pot and update-po
> targets and probably some other stuff.

Yeah, that sounds great.

> Thanks a lot for you help. It is really better now. I'd just love some
> comments about "cs" layouts.

Here you go. Perhaps others can comment further with their opinion
on the matter.

Cheers,
Danilo
prev next

Subscribe (FEED) Subscribe (DIGEST) Subscribe (INDEX) Unsubscribe Mail to Listmaster

Stalker Software, Inc.