From: Asmus Freytag (asmusf@ix.netcom.com)
Date: Tue Feb 01 2011 - 01:57:12 CST
Benjamin just saved me a longer discussion.
The proposal is to regularize the General Category for subscripted
characters along the same line as was done for superscripted characters.
As apparently there's not been a ill effect from fixing the superscript
characters, it should be possible to align the subscripted characters to
the same scheme.
That would have the benefit of removing an area of spurious difference
in treatment. Removing this is particularly beneficial, because of the
number of inter-linked properties affected, which makes it tricky for
the implementer to verify that the current difference in treatment is in
fact spurious and does not need to be reflected in the behavior of an
implementation in any way.
It also provides a cleaner set of precedents.
Finally, the relationship between GC=Lu/Ll/Lt/Lm and the other
properties ought to be captured better, perhaps in form of a small
matrix or table.
D135 defines cased in a way that is similar to a property stability
guarantee. Unless it is contemplated to revise D135 in the future, it
might be useful to document the fixed relationship between Cased and
it's constituent properties in a stability guarantee.
Further, it appears that case-ignorable can include "cased" characters
(Lm that are cased). This is surprising enough that it ought to be
spelled out (in a note).
A./
This archive was generated by hypermail 2.1.5 : Tue Feb 01 2011 - 02:00:33 CST