L2/05-238R

Source: Asmus Freytag
Date: August 23, 2005
Subject: Proposed change in properties for brackets

Murray and I were asked to propose properties for the 5 new brackets (shown below).

# email: asmusf@ix.netcom.com, 10 Aug 2005  13:38:33
#
# UnicodeData.txt entries for 5 new brackets. All should be lb=AL.

23DC;TOP PARENTHESIS;So;0;ON;;;;;N;;;;;
23DD;BOTTOM PARENTHESIS;So;0;ON;;;;;N;;;;;
23DE;TOP CURLY BRACKET;So;0;ON;;;;;N;;;;;
23DF;BOTTOM CURLY BRACKET;So;0;ON;;;;;N;;;;;
23E0;TOP TORTOISE SHELL BRAKCET;So;0;ON;;;;;N;;;;;


As part of our analysis we concluded that the *existing*  TOP SQUARE BRACKET and
BOTTOM SQUARE BRACKET  should have matching properties, as they are to be used
as the 6th and 7th member of this set of brackets.

Currently, they don't mirror. So that much is o.k. But they do have paired
punctuation properties (Ps/Pe) and Lb=OP/CL:

23B4;TOP SQUARE BRACKET;Ps;0;ON;;;;;N;;;;;
23B5;BOTTOM SQUARE BRACKET;Pe;0;ON;;;;;N;;;;;

23B4;OP # TOP SQUARE BRACKET
23B5;CL # BOTTOM SQUARE BRACKET


The properties should be:

23B4;TOP SQUARE BRACKET;So;0;ON;;;;;N;;;;;
23B5;BOTTOM SQUARE BRACKET;So;0;ON;;;;;N;;;;;

23B4;AL # TOP SQUARE BRACKET
23B5;AL # BOTTOM SQUARE BRACKET

Originally they were encoded together with 23B6 TOP AND BOTTOM SQUARE BRACKET
for use in terminal emulation. However, in that context, it is unlikely that
they are analysed as part of a linear text stream in a file or data stream.
Rather, it is more likely that they will be drawn as part of a single *column*
of text (each character on a line), or even as part of the whole screen buffer.

Furthermore, when used in that way, they are subject to a higher level protocol
which will effectively overried their GC and LB values. Therefore, it's more
important to give all 7 brackets a consistent set of properties.

Finally, the last item should match the two other members of *its* set,
therefore it should be

23B6;TOP AND BOTTOM SQUARE BRACKET;So;0;ON;;;;;N;;;;;
23B6;AL # TOP AND BOTTOM SQUARE BRACKET