Wm Seán Glen asked:
Couldn't one just embed the glyphs that aren't specified by Unicode along
with the text?
end quote
William Overton responded:
However, if one is using a plain unicode text file then one could not do
that embedding. The problem that then arises is that if one uses a code
such as U+E043 to represent a character that is not specified in Unicode,
with the U+E043 code being from the private use area, then if someone tries
to display that file, how will the software system displaying that file
know
which character to use.
...
Now, it might well be that the typesetting might be done using a package
that has formatting commands. Then there is no problem.
end quote
No, there are still *lots* of problems. Suppose your hypothetical U+E043
occurs between two letters that, oh, say, have different bidi
characteristic (e.g, between U+0065 and U+0628). How does the application
know the correct display order of the sequence? Answer: the application has
to be told the bidi properties of your U+E043. This is just one example of
many semantic properties that may need to be mutually understood.
Party A that places a PUA character into a file must document, somehow, a
number of semantic properties for each such PUA character for the benefit
of any Party B that wishes to "make sense" (other than in a degenerate
sense) of the characters in the file.
Bob Hallissy
This archive was generated by hypermail 2.1.2 : Fri Jul 06 2001 - 00:17:16 EDT