From cldr-users at unicode.org  Wed Apr 10 19:23:03 2019
From: cldr-users at unicode.org (Matthew Stuckwisch via CLDR-Users)
Date: Wed, 10 Apr 2019 20:23:03 -0400
Subject: Short number patterns
Message-ID: <99E857B5-EEFC-446E-9934-332087668426@softastur.org>

I am writing a module that uses CLDR data and am somewhat confused as to the correct method of interpreting CLDR number data.  Looking at English's short decimal formats, we have the following (trimmed for clarity)

<pattern type="1000" count="one">0K</pattern>
<pattern type="10000" count="one">00K</pattern>
<pattern type="100000" count="one">000K</pattern>

Given the number 12345, how does the result end up actually being 12K?  The number pattern 00 is for two digits, of course, but how is the number reduced to 12?  It can't be from significant digits, because those are defined by @ (and would result in trailing zeros), and it can be from maximum integer digits (which is neither set in the number pattern nor trims the smaller units).  Dividing by 10000 (from the type attribute) nets 1.2345, which for the pattern 00K should be 01K.

What have I been misinterpreting between the data and the standard or do the short forms have a different pattern syntax where 0 defines a sort of trimmed significant digits?




From cldr-users at unicode.org  Mon Apr 22 21:35:44 2019
From: cldr-users at unicode.org (Georges MURR via CLDR-Users)
Date: Tue, 23 Apr 2019 02:35:44 +0000 (UTC)
Subject: default numbering system for Arabic locales
References: <915405432.3286747.1555986944277.ref@mail.yahoo.com>
Message-ID: <915405432.3286747.1555986944277@mail.yahoo.com>

Hi,Is there a default numbering system for Arabic locales?? Is it the same for all countries? I am asking because I visited many Arabic government official websites and they all use Latin digits for both numbers and dates. All the Arabic newspapers websites that I visited do the same. Is there a default mapping between locale and numbering system in CLDR locale data?



Thanks,Georges
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://unicode.org/pipermail/cldr-users/attachments/20190423/eacdc11a/attachment.html>

From cldr-users at unicode.org  Mon Apr 22 23:43:46 2019
From: cldr-users at unicode.org (Markus Scherer via CLDR-Users)
Date: Mon, 22 Apr 2019 21:43:46 -0700
Subject: default numbering system for Arabic locales
In-Reply-To: <915405432.3286747.1555986944277@mail.yahoo.com>
References: <915405432.3286747.1555986944277.ref@mail.yahoo.com>
 <915405432.3286747.1555986944277@mail.yahoo.com>
Message-ID: <CAN49p6qNP-vQ=S7YR5==_1b8Rnwri0jeL07PE9y47R7GkLuBxQ@mail.gmail.com>

On Mon, Apr 22, 2019 at 7:37 PM Georges MURR via CLDR-Users <
cldr-users at unicode.org> wrote:

> Is there a default numbering system for Arabic locales?  Is it the same
> for all countries? I am asking because I visited many Arabic government
> official websites and they all use Latin digits for both numbers and dates.
> All the Arabic newspapers websites that I visited do the same. Is there a
> default mapping between locale and numbering system in CLDR locale data?
>

There is a default numbering system for all locales:
https://www.unicode.org/cldr/charts/latest/by_type/core_data.numbering_systems.html

For Arabic, the default is to use "native" (Arabic-script) digits, as is
customary in many countries like Egypt and Saudi Arabia.
Some regional variants, like Arabic in Morocco or Algeria, override this to
use ASCII digits.
For a few versions now, the data for all Arabic regional variants has
overrides for the region preference, so that this setting for Arabic itself
(without any regional variant) can be changed without affecting the region
locales.

Some of us have proposed to change the generic Arabic default to ASCII
digits because they are the most widely understood.

Best regards,
markus
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://unicode.org/pipermail/cldr-users/attachments/20190422/2986a10f/attachment.html>

From cldr-users at unicode.org  Tue Apr 23 01:21:49 2019
From: cldr-users at unicode.org (Richard Wordingham via CLDR-Users)
Date: Tue, 23 Apr 2019 07:21:49 +0100
Subject: default numbering system for Arabic locales
In-Reply-To: <CAN49p6qNP-vQ=S7YR5==_1b8Rnwri0jeL07PE9y47R7GkLuBxQ@mail.gmail.com>
References: <915405432.3286747.1555986944277.ref@mail.yahoo.com>
 <915405432.3286747.1555986944277@mail.yahoo.com>
 <CAN49p6qNP-vQ=S7YR5==_1b8Rnwri0jeL07PE9y47R7GkLuBxQ@mail.gmail.com>
Message-ID: <20190423072149.5a77ca70@JRWUBU2>

On Mon, 22 Apr 2019 21:43:46 -0700
Markus Scherer via CLDR-Users <cldr-users at unicode.org> wrote:

> On Mon, Apr 22, 2019 at 7:37 PM Georges MURR via CLDR-Users <
> cldr-users at unicode.org> wrote:
> 
> > Is there a default numbering system for Arabic locales?  Is it the
> > same for all countries? I am asking because I visited many Arabic
> > government official websites and they all use Latin digits for both
> > numbers and dates. All the Arabic newspapers websites that I
> > visited do the same. Is there a default mapping between locale and
> > numbering system in CLDR locale data? 
> 
> There is a default numbering system for all locales:
> https://www.unicode.org/cldr/charts/latest/by_type/core_data.numbering_systems.html
> 
> For Arabic, the default is to use "native" (Arabic-script) digits, as
> is customary in many countries like Egypt and Saudi Arabia.
> Some regional variants, like Arabic in Morocco or Algeria, override
> this to use ASCII digits.

But as I understand it, ASCII digits are the correct encoding* of the
Western Arabic digits!  The (near) Eastern Arabic digits (U+0660 etc.)
are not 'native' to the western part of the Arab world.

*Is there a Bidi test on printed text to distinguish, say, U+0662 from
U+06F2 EXTENDED ARABIC-INDIC DIGIT TWO?

Richard.

From cldr-users at unicode.org  Tue Apr 23 18:55:15 2019
From: cldr-users at unicode.org (Markus Scherer via CLDR-Users)
Date: Tue, 23 Apr 2019 16:55:15 -0700
Subject: default numbering system for Arabic locales
In-Reply-To: <20190423072149.5a77ca70@JRWUBU2>
References: <915405432.3286747.1555986944277.ref@mail.yahoo.com>
 <915405432.3286747.1555986944277@mail.yahoo.com>
 <CAN49p6qNP-vQ=S7YR5==_1b8Rnwri0jeL07PE9y47R7GkLuBxQ@mail.gmail.com>
 <20190423072149.5a77ca70@JRWUBU2>
Message-ID: <CAN49p6q7jBWdQ1OzhmdNf0aCFuA+80zFV24wA3frQb8Lxd-YtQ@mail.gmail.com>

On Mon, Apr 22, 2019 at 11:23 PM Richard Wordingham via CLDR-Users <
cldr-users at unicode.org> wrote:

> On Mon, 22 Apr 2019 21:43:46 -0700
> Markus Scherer via CLDR-Users <cldr-users at unicode.org> wrote:
> > For Arabic, the default is to use "native" (Arabic-script) digits, as
> > is customary in many countries like Egypt and Saudi Arabia.
> > Some regional variants, like Arabic in Morocco or Algeria, override
> > this to use ASCII digits.
>
> But as I understand it, ASCII digits are the correct encoding* of the
> Western Arabic digits!  The (near) Eastern Arabic digits (U+0660 etc.)
> are not 'native' to the western part of the Arab world.
>

The terminology is messy. I explicitly say "ASCII digits" vs. "native" (for
a given script) or "Arabic-script".

The U+06xx digits are native to the Arabic script, but you are right that
for Arabic in Morocco or Algeria they are not customary (as I said, and as
you can see in the data).

The U+06Fx digits are used for Persian and Urdu, not for the Arabic
*language*, but of course Persian & Urdu do use the Arabic *script*.

markus
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://unicode.org/pipermail/cldr-users/attachments/20190423/6d881d58/attachment.html>

From cldr-users at unicode.org  Tue Apr 23 20:04:10 2019
From: cldr-users at unicode.org (Richard Wordingham via CLDR-Users)
Date: Wed, 24 Apr 2019 02:04:10 +0100
Subject: default numbering system for Arabic locales
In-Reply-To: <CAN49p6q7jBWdQ1OzhmdNf0aCFuA+80zFV24wA3frQb8Lxd-YtQ@mail.gmail.com>
References: <915405432.3286747.1555986944277.ref@mail.yahoo.com>
 <915405432.3286747.1555986944277@mail.yahoo.com>
 <CAN49p6qNP-vQ=S7YR5==_1b8Rnwri0jeL07PE9y47R7GkLuBxQ@mail.gmail.com>
 <20190423072149.5a77ca70@JRWUBU2>
 <CAN49p6q7jBWdQ1OzhmdNf0aCFuA+80zFV24wA3frQb8Lxd-YtQ@mail.gmail.com>
Message-ID: <20190424020410.4ecb2df7@JRWUBU2>

On Tue, 23 Apr 2019 16:55:15 -0700
Markus Scherer via CLDR-Users <cldr-users at unicode.org> wrote:

> The terminology is messy. I explicitly say "ASCII digits" vs.
> "native" (for a given script) or "Arabic-script".
> 
> The U+06xx digits are native to the Arabic script, but you are right
> that for Arabic in Morocco or Algeria they are not customary (as I
> said, and as you can see in the data).

My claim is not only not customary now, but never were.

The decimal digits are younger than the Arabic and Latin scripts.

> The U+06Fx digits are used for Persian and Urdu, not for the Arabic
> *language*, but of course Persian & Urdu do use the Arabic *script*.

The difference between U+0662 and U+06F2 is that the former is bidi=AN
and the latter is bidi=EN.  My question was whether that difference was
anything more than a coding convention.  If there is a difference, then
perhaps the Maghribi forerunners of European digits need to be encoded
as bidi=AN.

Richard.