From: Tomas Frydrych (tomas@frydrych.uklinux.net)
Date: Wed Aug 14 2002 - 16:31:40 EDT
> The solution sounds feasible for actually performing the character
> remapping to screen. A larger and probably more interesting problem
> would be how to identify these smart-glyphs and differentiate between
> these and normal ones (the " inch marker, mismatched start/end quote
> pairs, ...
>
> Do you have any idea of how to feasibly do this? Might the unicode
> website help us?
There is no way to really differentiate between the two kinds of
glyphs, in terms of the information presented by the glyph codes
that problem is not deterministic, i.e., you cannot tell if "21" is a 21
in quotation marks, or 21 inches openning a quote, or is it?" -- you
can only tell that when you know not just the wider context but also
_meaning_ of the text in that context. Matching start and end quotes
is equally impossible for the very same reason, as long as you
cannot tell whether quote is really a quote, you cannot tell which
quotes should be paired.
Under the circumstances I see only two realistic options; (1) no
smart quotes at all. This is pretty much the case at the moment, and
I find it rather unsatisfactory. (2) Assume that all straight quotes are
quotes or mid-word apostorphies and let the user handle the rest
manually. If he does not like it, let him turn it off.
I think as soon as you start striving for something "better" than this,
i.e., you start assuming that you can work out when the quote is not
a quote (which you cannot), you will end up with behaviour which is
less predictable, and consquently more irritating for the user.
Anyway, that's what I think
Tomas
This archive was generated by hypermail 2.1.4 : Wed Aug 14 2002 - 16:36:56 EDT