From: Jukka K. Korpela (firstname.lastname@example.org)
Date: Thu Sep 21 2006 - 01:20:50 CDT
On Wed, 20 Sep 2006, Addison Phillips wrote:
> Locales that use spaces in digit groups generally use the regular
> non-breaking space character (U+00A0).
That's what there is in the CLDR data
( http://www.unicode.org/cldr/data/charts/by_type/number.symbol.html )
but I'm pretty sure that actual data almost universally contains just
normal spaces. Non-breakability and the amount of spacing are handled at
the styling and formatting level, if at all. This may slowly change in
computer-generated texts, as the utilization of CLDR grows.
> Less common spaces I would avoid: they
> may not translate well to legacy encodings or might not have glyphs available
> in specific fonts.
I wouldn't be so worried about conversions to legacy encodings when using
Unicode for new data. The other concerns are important. Fixed-width spaces
are generally very poorly supported in fonts, though the thin space might
be adequate in special cases (where you _know_ that the font in use
contains that character in a suitable appearance or the rendering engine
handles the thin space in a suitable way _and_ you can prevent line breaks
by some means).
The ideal character for plain text _would_ be a thin no-break space, but
it is neither widely supported nor sufficiently well defined. The Unicode
Standard does not describe its intended meaning and usage sufficiently
well, I would say - on the surface at least, it appears to be included for
specific use in a particular language.
> U+00A0 is generally available in most encodings and fonts and has the desired
> effect. Whether it is proportional or not depends, in large part, on the font
Typically, the no-break space has the same width as the space and it
preserves that width in formatting processes where the space may expand or
shrink (especially when producing text that is justified on both sides).
This may make it unsuitable for typographic reasons.
-- Jukka "Yucca" Korpela, http://www.cs.tut.fi/~jkorpela/
This archive was generated by hypermail 2.1.5 : Thu Sep 21 2006 - 01:23:13 CDT