Re: How to handle CJK ExtB, C, D?
----- 元のメッセージ -----
| Hi, all.
|
| As an i18n guy, you probably know what does "CJK ExtB, C, D" mean.
| If not, consult this page:
| http://en.wikipedia.org/wiki/CJK_Unified_Ideographs
|
| I have some test pages for CJK stuff:
| http://personal.ie.cuhk.edu.hk/~mx011/cjk.html#cjk
| https://en.wikipedia.org/wiki/List_of_CJK_Unified_Ideographs,_Extension_A
| http://personal.ie.cuhk.edu.hk/~mx011/extb.html#extb
| https://en.wikipedia.org/wiki/List_of_CJK_Unified_Ideographs,_Extension_C
| https://en.wikipedia.org/wiki/List_of_CJK_Unified_Ideographs,_Extension_D
|
| Tthe result of Fedora fresh install: only the first two regions are
| covered. (U+9FCC is a special case since it is added in Unicode 6.1)
|
| But our input methods do have ExtB, C, D stuff, thus leads to
| impedance mismatch.
| Example:
|
https://github.com/definite/ibus-table-chinese/blob/master/tables/array/a...
|
https://github.com/definite/ibus-table-chinese/blob/master/tables/cangjie...
|
| Well, it's all about fonts, right?
That may depends on applications you are talking about. but that's maybe true.
|
| wqy-zenhei convers up to ExtA.
| wqy-microhei doesn't even cover ExtA.
| WQY project seems dead for some time already.
|
| Recent release of droid is said to ExtA.
|
| hanazono covers up to ExtD, but:
| 1. It has Ming style (also called Song style), but people expect
| Black
| (Hei) style in modern desktop environments.
| 2. It should have Japanese variant of characters whenever applicable,
| may not be good enough for non-Japanese users.
Right. that said HanaMin might helps you to check if applications takes care of the
surrogate pairs properly. also that could gives you an idea to address your issue,
because, in fact this font is organized as two separate fonts, HanaMinA and HanaMinB,
which covers ExtA etc and ExtB, C, and D. the package contains fontconfig recipe to deal
with them as one font (the name `HanaMin' as the alias is came from what it was
originally named):
<match target="scan">
<test name="family">
<string>HanaMinA</string>
</test>
<edit name="family">
<string>HanaMin</string>
</edit>
<edit name="fullname">
<string>HanaMin</string>
</edit>
</match>
<match target="scan">
<test name="family">
<string>HanaMinB</string>
</test>
<edit name="family">
<string>HanaMin</string>
</edit>
<edit name="fullname">
<string>HanaMin</string>
</edit>
</match>
<alias binding="same">
<family>HanaMinA</family>
<accept>
<family>HanaMin</family>
</accept>
</alias>
<alias binding="same">
<family>HanaMinB</family>
<accept>
<family>HanaMin</family>
</accept>
</alias>
In packages, it may be not a good idea to do similar with different fonts but looks
similar variants though - e.g. when you find something out which covers Ext. B, C, and D
in various fonts, and looks acceptable for Chinese - you have an option to do like this in
your user configuration.
Please note that you have to re-run fc-cache after setting this up because it expects to
re-generate the fontconfig cache and vice versa. also it may not work if renderers
supports a single font only such as Xft.
Hope that helps,
| --
| i18n mailing list
| i18n(a)lists.fedoraproject.org
| https://admin.fedoraproject.org/mailman/listinfo/i18n