Re: WRAP_WORD with CJK

[Owen Taylor <otaylor redhat com>]

HideToshi Tajima <hidetoshi tajima eng sun com> writes:

> >> Is this a bug? If so, which module, pango, gtktextview or gtktextlayout,
> >> should be responsible for it? Have this been filed in bugzilla already?
> >
> >The default break implementation in pango/pango-break.c should handle
> >Japanese properly. (It implements algorithms from the Unicode standard
> >that take care of prohibited breaks, and so forth.) If it doesn't it
> >is a bug. I haven't it reported yet on bugzilla.
> 
> Prohibited breaks for punctuation marks, etc, work in general. They
> are wrapped down to the next line properly, but I don't see a difference
> b/w WRAP_CHAR and WRAP_WORD with Japanese.
> 
> For instance, Japanese text surrounding with opening and closing 
> quotes pair, opening and closing bracket pair, and so on, I expect the
> whole text should be wrapped down to the next line in WRAP_WORD.
> 
> In English, for instance, "foo", (bar), {barbar} are wrapped down as a whole
> in WRAP_WORD while wrapped as separate characters one by one in WRAP_CHARS.
> 
> Is it a valid bug?

I'd say not. In some sense, WRAP_WORD should correspond to the default
way of wrapping in any language. To my understanding, this is, for
Japanese, to allow breaks everywhere except where explicitely prohibited.

Otherwise, programs would have to be localized to specially enable
WRAP_CHAR for CJK locales, and WRAP_WORD for other locals, and most
programs would not get this right.

Also, how would you know that (bar) should be wrapped as a single
unit, but (this is a very long japanese text) should be broken?

Regards,
                                        Owen