Please accept belocs-locales-data 2.3.4-12

[barbier@linuxfr.org (Denis Barbier)]

This version is in unstable for 4 days, testing has 2.3.4-7, and I
will explain in this mail why I believe that this change is safe and
meets your criteria whereas interdiff reports that more than 55k lines
have changed in diff.gz files ;)

Several changes are due to moving hunks between patches under
debian/patches.  This package is Arch: all and has been built on a sarge
box, so I will give explanations about changes in binary packages instead
of source packages in order to focus on real changes:
   $ dpkg-deb -x belocs-locales-data_2.3.4-7_all.deb old
   $ dpkg-deb -x belocs-locales-data_2.3.4-12_all.deb new
   $ find old new -name \*.gz -exec gunzip {} \;

Changes can be summed up into 6 categories:
  1. Changes in date_fmt
  2. New tl_PH.ISO-8859-15 in SUPPORTED
  3. Backport of upstream CVS commits
  4. Other improvements in locales files
  5. Addition of first_weekday fields
  6. Improvements in GB18030 and UTF-8 charmaps.

I would like to push this version into testing because of the first
item.  It is very similar to translation updates because it changes
localized output of the 'date' command, and nothing else.

If other changes are too disruptive, I can make a new upload to unstable
with only some changes applied, please let me know.

Here are detailed explanations about differences.  First, let's check
locales files.  Changes in comment and first_weekday fields are not
considered:
  $ diff -uNr -I '^%' -I '^first_weekday' {old,new}/usr/share/i18n/locales
  cy_GB: s/<U006d>/<U006D>/ to quiet vim syntax highlighting
  el_GR: See http://sources.redhat.com/bugzilla/show_bug.cgi?id=588
  en_US: remove duplicate field
  fa_IR: Committed upstream into CVS
  fo_FO: Changes in d_t_fmt and date_fmt, see #307194
  fr_BE: Changes in date_fmt, and addition of yet unused fields
  fr_CA fr_CH fr_FR fr_LU: Likewise
  ku_TR: See http://sources.redhat.com/bugzilla/show_bug.cgi?id=870
         (with minor changes after private mail exchanges)
  pa_IN: See http://sources.redhat.com/bugzilla/show_bug.cgi?id=622
  sk_SK: Addition of first_weekday
  ta_IN: Cosmetics
  wo_SN: Update, see #306082
  zh_TW: Addition of first_weekday

Now, let's have a look at other changes:
  $ diff -uNr --exclude=locales old new 
  /usr/share/doc/belocs-locales-data/*: Packaging updates
  /usr/share/i18n/SUPPORTED: Addition of tl_PH.ISO-8859-15, see #304492
  /usr/share/i18n/charmaps/GB18030 /usr/share/i18n/charmaps/UTF-8:
      These are by far the most disruptive changes.  These files were
      modified to speed up locales generation (on my old box, compilation
      of zh_CN.GB18030 is 5 times faster, and UTF-8 locales is about 20%
      faster).  I carefully double checked these changes (put in
      2.3.4-10 uploaded on 2005-04-25) and did not encounter any problem
      so far.

To summarize, I believe that all these changes are suitable for sarge.
But if you are reluctant to push those charmaps changes into sarge, I
can make a new upload containing all other changes.
Thanks for your consideration.

Denis