NNBSP Impact

I am getting an error on the link below.
But this link, as provided by Richard works for fine for me ...  http://unicode.org/cldr/utility/character.jsp?a=202F

It would seem to me, offhand, that the main purpose of the 202F is as the name implies - is to provide space which cannot be broken - in the same way that 00A0-NoBreakSpace is used. I use the U+00A0 in my database to link compound words. I guess if I wanted that space between compound words to be narrower, I would use the NNBSP. So, it would seem that if this is the case, then we cannot change the NNBSP from being the white space that it is designed to be in other languages TO not being white_space in the language of our discussion - Mongolian.

The use of the NNBSP in Mongolian is mainly to provide space between the stem and the suffix AND between a suffix and a following suffix. I know of no other use for the NNBSP in Mongolian. It may be that we really do need another control character at U+180F.

Can anyone comment on the use of the NNBSP in languages other than Mongolian - such as French or Russian?

Greg


From: jrmt@almas.co.jp [mailto:jrmt@almas.co.jp]
Sent: Wednesday, July 8, 2015 7:39 AM
To: 'Badral S.'; public-i18n-mongolian@w3.org
Subject: RE: NNBSP-MVS Impact

Hi Greg,

Here I would like to provide Unicode property description of NNBSP.

NNBSP is included in Unicode white-space set according to following URL.
http://unicode.org/cldr/utility/list-unicodeset.jsp?a=[:White_Space=Yes<http://unicode.org/cldr/utility/list-unicodeset.jsp?a=%5b:White_Space=Yes>:]

What we need in Mongolian is non-whitespace categorized character to handle the suffixes.
Otherwise, all the suffixes will be separated from their main word in most of the applications.

Thanks and Best Regards,


Jirimutu
==========================================================
Almas Inc.
101-0021 601 Nitto-Bldg, 6-15-11, Soto-Kanda, Chiyoda-ku, Tokyo
E-Mail: jrmt@almas.co.jp<mailto:jrmt@almas.co.jp>   Mobile : 090-6174-6115
Phone : 03-5688-2081,   Fax : 03-5688-2082
http://www.almas.co.jp/   http://www.compiere-japan.com/
==========================================================



From: Badral S. [mailto:badral@bolorsoft.com]
Sent: Tuesday, July 7, 2015 11:31 PM
To: public-i18n-mongolian@w3.org<mailto:public-i18n-mongolian@w3.org>
Subject: Re: NNBSP-MVS Impact

Hi Greg,
Can you deliver some stabilisation problems for other languages as mentioned like Russian or French etc., because I didn't find something. Thus, I described it as best method to redefine that symbol. Otherwise, I wrote something contradictory.
If there exist stabilisation problems really, we need to forget NNBSP and we have to define new NNBSP control character to replace it's role. In this case I would also suggest 180F as Jirimutu mentioned. It should be named also "Mongolian Suffix Joiner".
I tend to think, we don't need to contact so many vendors to release it. May be Microsoft for Uniscribe, Harfbuzz for OSS world, SIL for ICU, Unicode for recommendation purposes. AAT and Graphite fonts include rendering rules itself. So, it's no problem for us font developers.

BTW: Bolorsoft font (MongolianScript) seems incorrect under http://r12a.github.io/scripts/mongolian/variants.html
May be it's very old version. I will provide our latest version asap.

regards,
Badral


On 07.07.2015 10:56, Greg Eck wrote:
Mr. Jirimutu,

I am not the one to answer about redefining the NNBSP since it is defined by the Unicode Consortium.  However, in reading the past posts from Badral and others, it appears that redefinition would be problematic at this point as it is tied to other languages such as French and Russian - stabilization would be the issue. Can someone else answer this question as stated below?

I can represent Dr. Quejingzhabu in his request that upper-level layout features in MS Word dealing with the NNBSP have been broken for some time. I think we had a discussion on this at least two years back. His bug report dealt with word-count across NNBSP-bounded suffixes, along with CTL-RIGHT/CTL-LEFT word skipping across the same NNBSP-bounded suffixes.

Greg

Are there other problems that other some of you have found in dealing with the current implementation of the NNBSP? Let's see if we can put together a list of problems found in the current NNBSP implementation ...
Spell-checking - Badral
Xxx - Mr. Jirimutu
Word-count - Prof. Quejingzhabu
Word-skip (CTL-RIGHT/LEFT) - Prof. Quejingzhabu
...



From: jrmt@almas.co.jp<mailto:jrmt@almas.co.jp> [mailto:jrmt@almas.co.jp]
Sent: Monday, July 6, 2015 9:17 AM
To: Greg Eck; public-i18n-mongolian@w3.org<mailto:public-i18n-mongolian@w3.org>
Cc: 'Behdad Esfahbod'
Subject: RE: NNBSP-MVS Impact

Hi Greg Eck,

Thank you very much. I appreciate your attention.
I will prepare my discussion paper for NNBSP include our AAT implementation logic.
We are discussing in our team how to propose for NNBSP. I will send our discussion paper in one or two days.

As Mr. Badral's mail, the NNBSP is the big problem in the Mongolian Encodings now.
Before explain our solution of NNBSP on AAT,
I would like to ask if is it possible to redefine the Suffix Joiner ?

If it is possible to redefine the Suffix Joiner, I would like to propose U+180F to be the Mongolian Suffix Joiner.
That will be pretty on the code structure and simple on the handle.

If it is impossible to redefine, we have to use NNBSP, I would like to participate into the discussion.

Thanks and Best Regards,

Jirimutu
==========================================================
Almas Inc.
101-0021 601 Nitto-Bldg, 6-15-11, Soto-Kanda, Chiyoda-ku, Tokyo
E-Mail: jrmt@almas.co.jp<mailto:jrmt@almas.co.jp>   Mobile : 090-6174-6115
Phone : 03-5688-2081,   Fax : 03-5688-2082
http://www.almas.co.jp/   http://www.compiere-japan.com/
==========================================================




From: Greg Eck [mailto:greck@postone.net]
Sent: Monday, July 6, 2015 1:17 AM
To: jrmt@almas.co.jp<mailto:jrmt@almas.co.jp>; public-i18n-mongolian@w3.org<mailto:public-i18n-mongolian@w3.org>
Cc: 'Behdad Esfahbod'
Subject: RE: NNBSP-MVS Impact

Mr. Jirimutu,
Thank you for the explanation on AAT's handling of the character preceding the MVS.
Please do let us know how the character preceding the NNBSP is handled also - that will be very helpful.
Greg


From: jrmt@almas.co.jp<mailto:jrmt@almas.co.jp> [mailto:jrmt@almas.co.jp]
Sent: Monday, July 6, 2015 12:04 AM
To: Greg Eck; public-i18n-mongolian@w3.org<mailto:public-i18n-mongolian@w3.org>
Cc: 'Behdad Esfahbod'
Subject: RE: NNBSP-MVS Impact

Hi Greg Eck,

I would like to send my some discussion about the MVS today.
I will follow up the NNBSP soon if I have any comment.

>It would be nice to find out whether Apple rendering systems follow suit.
>If anyone knows of an Apple engineer that we could ask,
>I will follow up on the matter. Or if there are other rendering systems >that we should consider, please bring them up.

We created the Apple AAT Font on our site and Apple handle this as an static machine stream.
In the case all of the rules we defined working well on Mac OS X and iOS system.
The MVS model processing on our AAT font is that

the string <MONG_INITIAL><MONG_MEDIAL>< MONG_LETTER ><MVS><U+1820 | U+1821> where MONG is the range U+1820 - U+18AA,

Our font will tag the MONG_LETTER as <fina>  only if the <MONG_LETTER> is one of <U+1828-n>, <U+182C-n>, <U+182D-g>,

<U+182E-m>, <U+182F-l>, <U+1830-s>, <U+1831-sh>, <U+1835-j>, <U+1836-y>, <U+1837-r>, <U+1838-w> as well as

<U+1823-o>, <U+1824-u>, <U+1825-oe>, <U+1826-ue>. Which Is the Mongolian MVS requested characters.

Note: Apple system site had one bug on the Mac OS X 10.8-10.10.2 and iOS 8.0-8.2

rdar://problem/18483089> REGRESSION: iOS8 did not able to correctly rendering Mongolian vowel separator(180E)

this bug had been fixed on Mac OS X 10.10.3 and iOS 8.3 now.


Thanks and Best Regards,


Jirimutu
==========================================================
Almas Inc.
101-0021 601 Nitto-Bldg, 6-15-11, Soto-Kanda, Chiyoda-ku, Tokyo
E-Mail: jrmt@almas.co.jp<mailto:jrmt@almas.co.jp>   Mobile : 090-6174-6115
Phone : 03-5688-2081,   Fax : 03-5688-2082
http://www.almas.co.jp/   http://www.compiere-japan.com/
==========================================================




From: Greg Eck [mailto:greck@postone.net]
Sent: Saturday, July 4, 2015 11:34 PM
To: public-i18n-mongolian@w3.org<mailto:public-i18n-mongolian@w3.org>
Cc: Behdad Esfahbod
Subject: RE: NNBSP-MVS Impact

Behdad Esfahbod, as the Harfbuzz designer,  was kind enough to answer my  questions of how Harfbuzz handles the MVS/NNBSP context. He confirmed my understanding that Harfbuzz follows the Microsoft Universal Shaping Engine in the following ... Thank you Behdad ...

>>>>>

Given the string <MONG_INITIAL><MONG_MEDIAL>< MONG_LETTER ><NNBSP><MONG_SUFFIX> where MONG is the range U+1820 - U+18AA, Harfbuzz applies the <fina> tag applied to MONG_LETTER

The same processing holds for <MONG_INITIAL><MONG_MEDIAL>< MONG_LETTER ><MVS><U+1820 | U+1821>
>>>>>

It would be nice to find out whether Apple rendering systems follow suit. If anyone knows of an Apple engineer that we could ask, I will follow up on the matter. Or if there are other rendering systems that we should consider, please bring them up.

The DS00 charts have been updated. I am also attaching two files that have been helpful to me in considering the range of usage of the MVS / NNBSP. DS04 deals with MVS usage. DS05 deals with NNBSP usage. Comments/corrections/questions are welcome.

Let's go on to consider six cases where the MVS/NNBSP affect the shaping behavior of the character immediately preceeding/following the MVS/NNBSP - U+1820, U+1828, U+182C, U+182D, U+1835, and U+1836.

Greg Eck



--

Badral Sanlig, Software architect

www.bolorsoft.com<http://www.bolorsoft.com> | www.badral.net<http://www.badral.net>

Bolorsoft LLC, Selbe Khotkhon 40/4 D2, District 11, Ulaanbaatar