et_EE locale - we still want to make it correct

[Meelis Roos <mroos at linux dot ee>]

I only now heard that there was some et_EE discussion in libc-alpha back
in January. The discussion is shown on the links below. The point of the
discussion was whether to change the definition of et_EE from ISO-8859-1
to ISO-8859-15.

The missing piece here is that the standard installation instructions to
get Estonian-language Linux environment include the instructions to
change et_EE locale to ISO-8859-15 charset by hand since glibc has it
wrong. I repeat, 8859-15 IS the de facto charset for et_EE - all
knowledgeable admins and major sites like the universities have used
8859-15 with plain et_EE for a long time without problems. So there's
actually no point of keeping the de jure et_EE in glibc - anyone who
wants to have full localization support changes it to 8859-15 by hand.
Or uses Solaris or smth that has it correct by default.

Also, ISO-8859-15 doesn't take away any character used in Estonian
language. It only adds the ones we use. So there's nothing that depends
on the wrong et_EE definition. We have tested it for years and found no
problems at all. Heck, I use it on all my computers for almost 2 years.

You say: "this creates incompatibilities.  Once a locale has a name its
encoding cannot be changed". What incompatibilities do you mean? We have
seen none. In addition, the fix would not actually change the definition
since there are two definitions being used in the wild and the fix would
just phase out the major source of the incorrect one.

We were very sad that it didn't make it into glibc last time it was
tried (some years ago) - the person who sent the original et_EE locale
data had left to another work and was not reachable to ACK the 8859-15
change. The current (since 2000) official Estonian standard is available
at http://www.eki.ee/itstandard/contents.html (it actually is a verbatim
copy of the official standard).

et_EE@euro is not what we need since we need our 2 missing letters (s
caron, z caron plus the same in uppercase) independently of euro
support.

et_EE@ISO-8859-15 would be correct but it creates confusion among users
and this is avoidable with the correct fix. Additionally, in my
experience, the locale@charset form of locale names unhides many bugs in
programs that don't strip out the @charset part and compare by full
strings.

This is the thread:

<URL:http://sources.redhat.com/ml/libc-alpha/2004-01/msg00016.html>,
<URL:http://sources.redhat.com/ml/libc-alpha/2004-01/msg00017.html>,
<URL:http://sources.redhat.com/ml/libc-alpha/2004-01/msg00025.html>,
<URL:http://sources.redhat.com/ml/libc-alpha/2004-01/msg00026.html> and
<URL:http://sources.redhat.com/ml/libc-alpha/2004-02/msg00196.html>

So PLEASE, fix et_EE to 8859-15 and save many-many admins from
having to do it by hand all the time!

-- 
Meelis Roos (mroos@linux.ee)