On 7/24/2015 2:59 AM, Frédéric Grosshans wrote:
Let me rephrase my remark in a less “stupid and dangerous” way.
If a LTR character has the BidiMirrored=No property, it may either
be mirrored or not when typeset in RTL, depending on other factors.
Specifically, the BidiMirrored property has not been specified for
ancient LTR scripts which are mirrored when RTL or boustrephodon,
like Italic, Runic, Archaic Greek, Archaic Latin, Egyptian
Hieroglyphs. Note that some RTL script, like Old North Arabian, are
mirrored when LTR.
We do want "BidiMirrorred=No" to be honored; for example for the arrows
and the ornate parens. And we do not want that to be overridden
The issue with the ancient scripts (or any script used to capture
paleographic texts) seems to be primarily with letter shapes, not
punctuation,
and further would apply only to unpaired forms.
A carefully written note would keep in scope all paired characters.
It would be nice if there was a property that covered them, but I'm afraid
that BidiMirroringGlyph does not cover the character pairs to use when
BidiMirrored=No and code points need to be substituted to get the RTL
layout correct. That kind of property would be useful for modern text,
e.g. to allow support for automatic re-layout from RTL to LTR and vice
versa for texts containing arrows.
Declaring all unpaired code points overridable "in certain contexts" or
"depending on other factors" might then work.
A./
