FYI, currently the maintainer doesn't want to show the modifier.
https://bugzilla.gnome.org/show_bug.cgi?id=536387#c38

Thanks for your comment, Takao.

Regarding the modifier, and reading the comment you pointed me to, I just want to make clear that I was not advocating for showing the @ on the UI. I too would rather prefer a more "human" description. In the case of the Valencian variant (ca@valencia), both 'ca' and 'ca@valencia' use the same script, so I think it makes sense to show both entries.

Currently we've got the following identical entries, which make it impossible for users to tell the two locales apart:

Catalan (Spain) <- That's 'ca'
Catalan (Spain) <- That's 'ca@valencia'

A proposal would be to use a different name for the language:

Catalan (Spain)           <- That's 'ca'
Southern Catalan (Spain)  <- That'd be 'ca@valencia'

In any case, it seems to me that there might not be a solution that applies to all "@ locales".

(Also note that I've only used the "Southern Catalan" name by way of an example - if the implementation would end up using an alternative language name, this name should be discussed within the Catalan community)

Currently GDM shows UTF-8 locales only in the language list by default.

My understanding is, currently "ca" locale is not shown in GDM by default since it fails setlocale, export LANG="ca", until you run localedef by manual.
I think currently GDM shows "ca_ES.UTF-8" locale but "ca_ES.UTF-8@valencia" is not shown because setlocale is failed.

So I think you need to create ca_ES.UTF-8@valencia by manual with localedef at first.

Probably your idea would be nice.
The string "Catalan" is extracted by /usr/share/xml/iso-codes/iso_639.xml .
And the translation is extracted by /usr/share/locale/ca/LC_MESSAGES/iso_639.mo .

e.g.
% env LANG=ca_ES.utf8 gettext -d iso_639 "Catalan; Valencian"
català; valencià

The string is shown on GDM.
I think if you want to distinguish the GUI string between ca_ES.utf8 and ca_ES.utf8@valencia, one idea is to deliver iso_639.mo for ca@valencia .

e.g.
msgid "Catalan; Valencian"
msgstr "Català South; Valencià"

# I recommend "C" is first because the name is sorted on the GDM language list.

Then probably I think GDM could be enhanced to show the modifier locale in case the iso translations are different.

So your issue might be iso_codes l10n prior to GDM.

It is the work of GDM to show locale properly! And it should.

Finally, @variations are GNU extension and have no bearing to ISO 639.

Just provide the separate translations for common variations as @latin or @euro and show other variants unmodified.

For example,

be_BY.UTF-8@latin => "Belorussian (Belarus, Latin script)"

ru_UA.UTF-8@foobar => "Russian (Unkraine, foobar)"

Created attachment 176440 [details]
Now it's getting better...

CONFIRM it.
FIX it.

DON"T procrustinate.

As I noted in Comment #3, currently I think it's good to fix in iso-codes.
I think that translation fix could resolve your problem.

The maintainer would not like to show the modifiers locales on linux.
https://bugzilla.gnome.org/show_bug.cgi?id=536387#c35

> Finally, @variations are GNU extension and have no bearing to ISO 639.


I don't think it's GNU extension.
http://pubs.opengroup.org/onlinepubs/007908799/xbd/envvar.html

> I think that translation fix could resolve your problem.

While this is a workaround, it is a bit of a hack.

The iso-codes translations only contain language names, and no indication of the variant.

So in the particular case of variants that use the same script as the main language code, you'd end up with two identical entries in the GDM locale chooser, without being able to know which one is which, unless you translate the name of the language differently in iso-codes, which just feels wrong.

Maybe I think the maintainer concerns the language list is redundant if we just show the variant locales.
It may be good to think another GUI.
e.g.

[Add variant ▶] [ * None]
                [ o Latin (@latin) ]
                [ o Euro (@euro) ]
                [ o Cyrillic (@cyrillic) ]
                [ o Iqtelif (@iqtelif) ]
                [ o Abegede (@abegede) ]
                [ o Devanagari (@devanagari) ]
                [ o Saaho (@saaho) ]

At the moment, you could add modifiers in /var/lib/gdm/Seat1/.gconf/apps/gdm/simple-greeter/%gconf.xml directly or add LC_TIME=be_BY.utf8@latin in $HOME/.profile

However if you think about LC_TIME, I wonder adding %EY or %Oy might resolve the issue besides %y.
I'm a bit busy for GNOME3.

Hi Takao,

Thanks for your comment. I'm not too sure I understand your proposal, may I ask you to elaborate a bit on it?

Also I cannot really grasp the reference to LC_TIME now.

Thanks!

(In reply to comment #9)
> may I ask you to elaborate a bit on it?

I mean the default enables the "None" toggle button on the additional GUI and it won't add any variants for locales.
If the user choose "Latin" on the GUI, it will add the selected locale with "@latin", e.g. be_BY.utf8@latin .
It's just an idea.

> Also I cannot really grasp the reference to LC_TIME now.

You could refer the man page of date and strftime. You could use the %Y by default but also can use %EY for the alternative format.

As from the Ubuntu GDM version 2.32.0-0ubuntu7, where the downstream bug was fixed, the list of languages in GDM's language chooser may look like this:

Catalan
Catalan - valencia
English
English (Australia)
English (Canada)
English (New Zeeland)
English (United Kingdom)
English (United States)
German
Swedish

It shows available translations instead of available locales. Since
- Catalan is currently one of the installed languages on my box,
- there is a separate ca@valencia translation, and
- a matching locale is available,
a separate "Catalan - valencia" item is included in the list.

https://launchpad.net/ubuntu/+source/gdm/2.32.0-0ubuntu7

We no longer have this selector in GDM (it was moved to within the session) so closing...