[Bug localedata/17426] Indian locales: set the correct date format

["digitalfreak at lingonborough dot com" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=17426

Rafal Luzynski <digitalfreak at lingonborough dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|en_IN: set date format      |Indian locales: set the
                   |                            |correct date format

--- Comment #9 from Rafal Luzynski <digitalfreak at lingonborough dot com> ---
This needs further work so I change the bug summary.


Statistics
==========

Glibc currently supports 34 Indian locales (see the incomplete list in comment
7).  Only 19 of them are also supported by CLDR.  For others we must figure out
the correct date format based on other similar locales.

There is one locale (bo_IN) which uses "copy..." for LC_TIME instead of
providing its own content and one locale (ccp_IN) which is supported by CLDR
but not by glibc.  I will ignore those two.


Current state
=============

The most popular value of d_fmt in Indian locales is "%A %d %b %Y".  It is
provided by 25 locales: anp_IN, bhb_IN, bho_IN, bn_IN, brx_IN, doi_IN, gu_IN,
hi_IN, hne_IN, kn_IN, kok_IN, ks_IN, ks_IN@devanagari, mag_IN, mai_IN, mni_IN,
mr_IN, pa_IN, raj_IN, sa_IN, sat_IN, sd_IN, sd_IN@devanagari, tcy_IN, ur_IN.

The second most popular is "%A %d %B %Y".  Supported by 4 locales: ar_IN,
mjw_IN, ml_IN, ta_IN.  It has also been supported by the 5th locale: en_IN but
it has been changed to "%d/%m/%y" just days ago.  Most probably your computers
still provide the old date format for en_IN.

as_IN uses "%e-%m-%Y".

or_IN uses "%Od-%Om-%Oy".  It seems to me that someone made an effort to adapt
the locale to the actual needs of the local community.  I will not touch this.

te_IN uses "%B %d %A %Y".  This looks like a typo (Month dd Weekday yyyy) and
I'm pretty sure someone actually meant "%A %d %B %Y".


Proposed changes
================

For those locales which currently use "%A %d %b %Y" or "%A %d %B %Y" CLDR
provides "d/M/yy" (day number, as many digits as required - not padded, not
truncated; month number, as many digits as required; year number, truncated to
2 digits).  Glibc equivalent is: "%-d/%-m/%y". Locales: bn_IN, gu_IN, hi_IN,
kn_IN, ml_IN, mr_IN, pa_IN, ta_IN, ur_IN.  (9 locales)

Consequently I think that these locales which are not supported by CLDR but
currently provide the same value of d_fmt should accept the same change:
anp_IN, ar_IN, bhb_IN, bho_IN, doi_IN, hne_IN, mag_IN, mai_IN, mjw_IN, mni_IN,
raj_IN, sat_IN, sd_IN, sd_IN@devanagari, tcy_IN.  (15 locales)

However, it may be disputable if actual Indian users want the month number to
be zero-padded or not.  That means, should today date be formatted as "30/8/18"
and Sept 1st as "1/9/18" or maybe rather "30/08/18" and "1/09/18" or even
"30/08/18" and "01/09/18".

Note that ur_IN is in this group and ur_PK currently provides "%d/%m/%Y"
("30/08/2018" and "01/09/2018").

For as_IN CLDR suggests "d-M-y" which should translate to "%-d-%-m-%y" or
"%-e-%-m-%y".  Current value is "%e-%m-%Y" (zero padded month number, 4 digits
year).  Maybe leave unchanged?

For brx_IN and ks_IN which use "%A %d %b %Y" CLDR provides "M/d/yy" which
translates to "%-m/%-d/%y".  Consequently I think that ks_IN@devanagari which
is not supported by CLDR should have the same value.  (3 locales)

For kok_IN which currently uses "%A %d %b %Y" CLDR provides "d-M-yy" which
translates to "%-d-%-m-%y".

For sa_IN which currently uses "%A %d %b %Y" CLDR provides "d-MM-yy" which
translates to "%-d-%m-%y".  (Difference: in sa_IN month number should be
zero-padded to 2 digits, in kok_IN should not).

For te_IN which currently uses "%B %d %A %Y" (a typo IMHO) CLDR provides
"dd-MM-yy" which translates to "%d-%m-%y".

en_IN has been already updated.

Thoughts?

-- 
You are receiving this mail because:
You are on the CC list for the bug.