Hi Philippe,
Thanks for your opinion. The examples I've just checked looked like
"zero-width or almost so" (Friedl) and "very narrow" (in a Japanese
dictionary, ISBN 978-4-7674-2315-9) to me. I am pretty sure I've seen
zero-width with either kana or romaji, just can't find a reference right
now. I think there are use instances. What "should" be done is also a
good question, and perhaps the answer isn't as obvious as I thought, but
these considerations make me wonder whether a font is really free to
make a choice about zero-width or not for any given Unicode symbol. If
not, I'm wondering whether Unicode could/should/does make that binary
decision. And in either case I'm wondering for /which/ characters in
Unicode zero-width is an option (or obligatory).
Stephan
On 6/6/2012 4:52 PM, Philippe Verdy wrote:
They are not supposed to be zero-width. But kerning may still reduce
their width contextually, as long as there's no collision.
For use in Japanese as mora indicators, given the size of glyphs used
in Japanese (including Kanji characters) I don't think they should
even be kerned with them, but should behave like the existing
ideographic square quotation marks (and probably in this case, even
their advance width would be extended to fit the alignment of
ideographic squares, if they are used with an ideographic character
encoded immmediately near them on their inner side, and aligned to the
half-width box if their are used with an half-width character on their
inner side.
Fonts can contextually make these transformations of glyphs for Asian texts.
Outside of these contexts, these characters should behave and be
rendered like the usual [square brackets], with the same metrics and
similar kerning rules. In monospaced fonts, these characters should
advance by one cell and should be centered horizontally. They should
also include an empty side-bearing gap between the horizontal stroke
and the side of the cell (i.e. no connection/collision should be
possible with characters encoded on the inner side, unless these
characters are explicitly connecting outside of their own advance box,
a basic macron diacritic not creating a collision because it already
widdens the base character encoded with them to provide a minimum side
bearing and no collision with surrounding glyphs; font authors and
renderers will do their best to avoid these collisions and unexpected
joinings)
2012/6/7 Stephan Stiller<[email protected]>:
Hi,
I am excited to see the half-bracket symbols (U+2E22 -> U+2E25) in Unicode.
I've been waiting for them. I assume that they're meant for use cases such
as:
in certain Japanese dictionaries to indicate pitch accent (U+2E22 and U+2E23
would surround the high-register "mora")
for alternate quoting, such as Jeffrey Friedl's usage in his "Mastering
Regular Expressions" book to denote the beginning (U+2E22) and end (U+2E25)
of regexes
In both these two use cases that I'm aware of, the symbols are zero-width,
or casually indistinguishable from zero-width. Is there a place in the
standard to indicate that this is the recommended way for fonts to implement
these symbols?
