Re: CJK glyph selection

On Tue, Dec 20, 2011 at 1:44 PM, Kang-Hao (Kenny) Lu <
kennyluck@csail.mit.edu> wrote:

> cjk-glyph-sane.png (zh-TW on zh-TW Win7)
>
> (default): zh-TW
> ja: ja
> zh-TW: zh-TW
> zh-CN: zh-TW
> en: zh-TW
>
> cjk-glyph-insane.png (ja on zh-TW Win7)
>
> (default): zh-TW
> ja: ja
> zh-TW: zh-TW
> zh-CN: zh-CN
> en: ja
>
> (This is leaking OS-locale I guess, which seems bad.)
>

That's too bad.  I had some minute hope that there might be a sane
precedent for defaulting to Japanese fonts for non-@lang-marked UTF-8.  (It
wasn't much of a hope, but enough that it was worth looking at more
closely.)

Leaking the locale (codepage, for Windows) is bad, but that's exactly what
IE has always done.

It definitely seems bad if Firefox ends up depending on *both* the user's
locale and which language of Firefox installer that the user used.  That's
adding a second axis; a 2D matrix of possible results for just one
browser.  I'd be tempted to call that worse than IE's behavior, which is at
least one-dimensional.

cjk-glyph-intermide.png (zh-TW on ja Mac OSX and zh-TW on ja Mac OSX set
> to zh-TW)
>
> (default): ja
> ja: ja
> zh-TW: zh-TW
> zh-CN: zh-CN
> en-US: zh-TW
>
> This is matching Henri's statement as in [2] but I can't produce similar
> result on Win7 as seen by zewt[3], which looks like
>
> (default): ja
> ja: ja
> zh-TW: zh-TW
> zh-CN: zh-CN
> en-US: ja
>
> Having en-US different form the default seems odd by the way. Perhaps I
> am missing something...
>

I agree in principle, but maybe not in practice.  Even if the default (no
@lang at all) must have weird magic for legacy content, that might not be
true for lang=en.  In that case, it might be possible to define non-CJK
languages as always choosing Japanese glyphs.  That would at least
compartmentalize the annoying locale- and user-language-sensitive behavior
to pages with no @lang at all.  I have no data to guess whether legacy
content would permit this or not.

-- 
Glenn Maynard