[XeTeX] XeTeX 0.9999.0 released
zdenek.wagner at gmail.com
Mon Mar 18 12:18:52 CET 2013
2013/3/18 Khaled Hosny <khaledhosny at eglug.org>:
> On Mon, Mar 18, 2013 at 11:11:18AM +0100, Zdenek Wagner wrote:
>> 2013/3/16 Khaled Hosny <khaledhosny at eglug.org>:
>> > Hi,
>> > I can not really tell if this is a bug or a feature, bug HarfBuzz
>> > normalization is independent of input normalization to technically the
>> > input text is not normalised, but it makes little difference with the
>> > PDF output.
>> > If you still think this is an issue, please raise it again on HarfBuzz
>> > mailing list since your old message went unanswered.
>> I know nothing about Chinese but as I know, in Devanagari characters
>> with nukta as QA, ZA, FA (and a few others) exist as glyphs with their
>> own codepoints but can also be entered as KA, JA, PHA with combining
>> nukta. As I read somewhere, some fonts contain QA, ZA, FA, some fonts
>> do not. If JA + nukta is normalised to ZA but ZA does not exist in the
>> font, the character could not be rendered. NFD may be equally bad, if
>> ZA is decomposed to JA + nukta and the font does not contain nukta,
>> the character could not be rendered. So it may be an issue but I do
>> not know whether such problems may occur in nowadays' fonts. Similar
>> problem may occur even in latin scripts with combining diacritical
>> marks. The accented glyphs need not exists. Nowadays' fonts contain
>> all accented characters needed for languages as Czech, Slovak, Polish,
>> French but I do not know how widely Vietnamese is supported.
> HarfBuzz takes account of what glyphs exist in the font when choosing
> the normalization strategy, so such issue should not occur, if it does
> then it is a bug and let us know about it.
Thank you for explanation. My note was just theoretical, all fonts I
use (both commercial and free) contain all glyphs, so for me
personally NFC is no problem.
> Subscriptions, Archive, and List information, etc.:
More information about the XeTeX