On Fri, 2007-12-21 at 09:12 +0900, mpsuzuki@hiroshima-u.ac.jp wrote:

...

Arabic is like Japanese in that regard, no difference. I actually see that coming, should have clarified. By unexpected, I mean it's not the most likely event. Japanese text coming is more expected.

That said, we don't have that in issue as much in Arabic because it's considered bad writing to start an Arabic/Persian paragraph with an English word written in Latin. It also screws bidirectional code in Pango and you end up with a left-to-right paragraph (because that's what it looks like from your text), so people just avoid it.

I see. Hearing "people just avoid it" is quite interesting.

As I said, it's not just technical. It's bad style to start a Persian/Arabic paragraph with a Latin word.

...

But then when rendering a Japanese only text, all the punctuation marks will be rendered using a different font! Now imagine that in a monospace text, with bitmap Japanese font and non-bitmap punctuation font.

Yes. Do you think it's worse than contextual font switching?

I think rendering single-script text correctly is more important, yes. If you have plain text in one script, it should only use the preferred font of that script. Can't compromise here.

I don't think so. But it's because my fonts have varied/inconsistent baselines and heights (and their inconsistency makes the contextual font switching quite ugly), so my disagree is not so strong at present.

Anyway, your mention on bidi reminded me that binding a fixed font to COMMON characters may confuse bidi glyph shaping of punctuation. If so, it would be problematic and binding should be disabled even if it's possible. Oops.

No, bidi reordering is done independent of font selection. Those are completely separate processes.

...

...

...

I have two suggestions for what you can do that may achieve better results for you.

• Run under LC_LANG=en_US LC_MESSAGES=ja_JA

• Choose a non-generic font family in gedit. That is, something other

than Sans, Sans-serif, and Monospace.

Oops, it's too application specific...

No. Give it a try. It should have the effect you asked for. All punctuation should be chosen from the non-generic font you choose. I said do it in gedit just to test, otherwise it's nothing specific to gedit, that's how fontconfig works.

OK, I will try to setup ~/.fonts.conf.

I don't think that would do it. Just set it in gnome-font-properties.

It seems that my request (binding a same font to COMMON character, at least in Latin & CJK context) can be realized by it

Not exactly. Hardcoding a font in your fontconfig config to always return a certain font as the first font is not a good idea, and is actually what started this thread at the beginning.

• so it's off-topic to this list? Should I move to fontconfig?

I don't think forcing to use the same font for COMMON characters is really a solution. The simplest solution for the case you showed is to use a font that has both Japanese and Latin glyphs (plus all the punctuation). Again, what started this thread was that the CJK font had Latin glyphs, but crappy ones.

Anyway, thank you for enlightening me.

Regards, mpsuzuki