Bidi Brackets for Dummies

Eli Zaretskii eliz at gnu.org
Fri Apr 25 03:54:32 CDT 2014


> From: "Whistler, Ken" <ken.whistler at sap.com>
> Date: Thu, 24 Apr 2014 21:16:38 +0000
> Cc: "Whistler, Ken" <ken.whistler at sap.com>
> 
> Given the incredible level of interest shown on this list during
> the last week, I am glad that I can finally announce the publication
> of Bidi Brackets for Dummies:
> 
> http://www.unicode.org/notes/tr39/

Thanks.

I found one typo:

  Now that we have comes to grips with the fact
                   ^^^^^

I also have a couple of questions about matching the canonical
equivalents of the opening bracket:

1. Some characters have the decomposition mapping that starts with a
tag, such as "<small>" or "<vertical>".  Since (according to UAX#44,
paragraph 5.7.3) these indicate a compatibility mapping, not a
canonical mapping, I understand that they are not relevant for the
purposes of the BPA.  IOW, U+3008 and U+FE3F _cannot_ form a bracket
pair, even though U+FE3F has "<vertical> 3008" as its decomposition
mapping.  Is that understanding correct?

2. Why aren't pairs with canonically equivalent characters, such as
these:

  2329; 3009; o # LEFT-POINTING ANGLE BRACKET
  232A; 3008; c # RIGHT-POINTING ANGLE BRACKET
  3008; 232A; o # LEFT ANGLE BRACKET
  3009; 2329; c # RIGHT ANGLE BRACKET

in BidiBrackets.txt?  If they were, that file could serve as a single
source of information for deciding on bracket pairing; as things are,
the implementation of the BPA must access other Unicode properties to
do its job (unless the text representation is already decomposed such
that each character is represented by its canonical equivalent --
which is, of course, a complication for text editors).



More information about the Unicode mailing list