Re: getting unicode chars to show on Windows

[Jason Rumney]

On Aug 26, 7:35 pm, Xah Lee <xah...@gmail.com> wrote:

> humm... it show for you Eli out of the box. Strange.

Some fonts, particularly CJK ones which tend to cover a substantial
part of the BMP but have many gaps, seem to get in the way of (nearly)
full coverage fonts like Arial Unicode MS by default.

> am guessing the problem really needs to be solved by some sort of font
> substitution. Browsers apparantly are all doing it and perfectly on my
> machine since they show all chars without user needing to set fonts.

I don't know how the browsers are doing it, perhaps they come with
predefined lists of fonts that are commonly installed on Windows to
cover different ranges.

> i guess in emacs that is fontset? I'm not sure if it is just fontset,
> or emacs also calls the OS's font api to complete part of the
> display...

Yes, in Emacs that is a fontset, and the difference between what
people are reporting is caused by the fact that the default fontset
defines rules that are based on information other than the font name
by default.  If you know what fonts best support different characters
on your system, I'd recommend redefining the default fontset to use
your preferred fonts using (set-fontset-font t ... ...). There are
many examples of using this function in lisp/international/fontset.el,
along with a list of scripts that Emacs recognizes as the third
argument (in script-representative-chars); another simple one below:

(set-fontset-font t 'phonetic "Lucida Sans Unicode")