Re: Updating/adding locale for Ethiopia and Eritrea in GNU libc.

[Petter Reinholdtsen <pere at hungry dot com>]

Sorry for my late reply.  It has been a while since I've worked on
glibc.

[Daniel Yacob]
> The collection of new and updated locales for East Africa mentioned here
> earlier is now available in the following package:
> 
>   http://yacob.org/glibc-yeha.tar.gz

I fetched this, and tested the sorting of am_ET and gez_ET.  I ran
into a build problem with gez_ET, and had to comment out the lang_lib
keyword.  I'm not sure why.

I got a few comments.  Today, I learned new things about the
LC_IDENTIFICATION section.  The 'language' and 'territory' should use
the ISO-639 and ISO-3166 codes, not the full name.  And the 'category'
should reference the standard describing the format.  Almost all
locales have incorrect content in the category field.  This is
probably a good set of values:

  category  "i18n:2000";LC_IDENTIFICATION
  category  "i18n:2000";LC_CTYPE
  category  "i18n:2000";LC_COLLATE
  category  "i18n:2000";LC_TIME
  category  "posix:1993";LC_NUMERIC
  category  "i18n:2000";LC_MONETARY
  category  "i18n:2000";LC_MESSAGES
  category  "i18n:2000";LC_PAPER
  category  "i18n:2000";LC_NAME
  category  "i18n:2000";LC_ADDRESS
  category  "i18n:2000";LC_TELEPHONE

I notice the use of collate_et and translit_et.  If I understand the
glibc maintainers correctly, they want to avoid such include files,
and would rather store the content in a regular locale.  The 'copy'
statement should do the right thing.  This also make it easier to test
the sorting order using the current test framework.  If you just just
pick one of the locales, and include the content of collate_et and
translit_et in it, the locale can be copied into the other locales
just like collate_et and translit_et.