L2/09-197

Title:  Line Break Property Adjustments

Author: Ken Whistler

Date:   May 7, 2009

Action: For Consideration by the UTC


In the process of developing the draft data files for
the Unicode 5.2 beta of the UCD, I turned up a few
small anomalies for existing Line_Break property
assignments.

I suggest that the UTC consider and correct these
assignments for Unicode 5.2.


1. Javanese and Balinese "windu" characters.

The "windu" characters are ring punctuation, occurring
both in Balinese and in Javanese. The Amendment 6
documentation for Javanese suggests that this
punctuation mark is not actually a particular
break opportunity (lb=BA), and is often flanked
by other symbols/punctuation marks. Accordingly,
for Unicode 5.2 I have given it the default lb=AL
value.

The corresponding windu in Balinese apparently has
much the same behavior. It was encoded in Unicode 5.0
and given lb=BA, based on less clear information.
I suggest that the properties for the two characters
be aligned better now by updating the Line_Break
value for the Balinese windu:

   U+1B5C BALINESE WINDU: lb=BA-->lb=AL
   
   
2. North Indian currency signs

The addition in Amendment 6 of the North Indian
number forms and the Bengali ganda sign (a prefixed
currency sign) have highlighted some problems of
interpretation of Bengali signs that were encoded
long ago, before all the analysis and examples of
their usage became available. 

In particular, the Bengali rupee signs have been
lb=PR, which is wrong. Rupee signs are *post*fixed
currency signs. So I suggest the following changes:

   U+09F2 BENGALI RUPEE MARK: lb=PR-->lb=PO
   U+09F3 BENGALI RUPEE SIGN: lb=PR-->lb=PO
   
The Bengali currency denominator is actually another
of these postfixed currency signs, so it should
also have its Line_Break property value updated:

   U+09F9 BENGALI CURRENCY DENOMINATOR SIXTEEN: lb=AL-->lb=PO