L2/07-194

Date/Time:    Tue May 15 00:01:07 CDT 2007
Contact:      yacob@geez.org
Name:         Daniel Yacob
Report Type:  Public Review Issue
Opt Subject:  PR #96 Comment

This comment is available as an MS Word document and I 
can submit it via email if requested.

Ethiopic Wordspace (U+1361) in Identifiers

This is a multipart comment concerning the use of the Ethiopic 
wordspace character in identifiers. The Ethiopic letter symbol 
is presently in the “Punctuation Other” class.  It is proposed 
here that the symbol be given a joiner property within identifiers.
A “joiner property” to the same degree that hyphen and underscore 
become joining characters in identifiers. The symbol serves the 
same fundamental purpose in providing a non-intrusive visual word boundary.

As a joining symbol in the identifier context, it is further proposed 
that the Ethiopic wordspace be optional. When present it will be 
ignored and string interpreters may elide the symbol from an 
identifier. The intention here is to avoid potential phishing 
problems that may occur in IDN usages. Rather than add the elision 
rule as a special rule for IDNs, it is recommended for all identifiers.
For example:

  የኢትዮጵያ፡ንግድ፡ባንክ.com
  የኢትዮጵያንግድ፡ባንክ.com
  የኢትዮጵያ፡ንግድባንክ.com
  የኢትዮጵያንግድባንክ.com
  የ፡ኢ፡ት፡ዮ፡ጵ፡ያ፡ን፡ግ፡ድ፡ባ፡ን፡ክ.com

would all be equivalent, referring to the same domain and requiring 
only a single registration. Given the syllabic nature of the writing 
system, occasions where the elision rule could lead a single concatenation 
of independent words sets should be very rare I.E. የኢትዮጵያ፡ንግድ፡ባንክ should 
be the only meaningful sequence of strings within የኢትዮጵያንግድባንክ and it 
should be very rare that another sequence, like የኢትዮጵያን፡ግድባንክ, would also 
be meaningful.


Validity Pattern:
The wordspace must both be preceded and followed by a non-punctuation 
(i.e. letter or numeric) Ethiopic letter symbol:
/[\p{Ethiopic} -\p{Punctuation Symbol}]፡?[\p{Ethiopic} -\p{Punctuation 
Symbol}]/


IDN Example:
hxxp://www.የኢትዮጵያንግድባንክ.com/         hard to read
hxxp://www.የኢትዮጵያ-ንግድ-ባንክ.com/       minor strain to read
hxxp://www.የኢትዮጵያ፡ንግድ፡ባንክ.com/      easiest, most natural to read


Programming Language Example:
$አዲስስም = …    # hard to read
$አዲስ_ስም = …   # looks bizarre
$አዲስ፡ስም = …   # easiest, most natural to read