From: <unifoundry_at_unifoundry.com>
Date: Thu, 12 Mar 2009 11:34:35 -0700

On Thu, March 12, 2009 8:53 am, Jonas Häggqvist
<rasher(at)rasher(dot)dk> wrote:
> unifoundry(at)unifoundry(dot)com wrote:
>>
>> On the licensing, it is distributed under GPL v2 with font embedding
>> exception because I included thousands of CJK ideographs from
>> Qianqian Fang's Wen Quan Yi font, and that is the licensing under which
>> Wen Quan Yi is distributed. It is a free license, so there shouldn't be
>> any worries in including the font with Rockbox.
>
> This should be noted somewhere, but I can't find anything in our sources
> to indicate any license for Unifont. Does anyone else know where we have
> this kind of info put down (if anywhere)?

It is mentioned in the source tarball (which was my main focus while
scrambling to meet the Debian 5.0 release deadline), but could be stated
better. And my main focus then was the TrueType version, because the
BDF version wasn't properly displaying combining characters on xterm. I
have come to learn that is a deficiency of xterm, not of the BDF font.
I will update the web page shortly to make GPL v2 licensing for anything
using Wen Quan Yi glyphs clearer.

The licensing was mentioned on Rockbox IRC at 00:38:

     http://www.rockbox.org/irc/log-20080826

It is also mentioned in the discussion of the Wen Quan Yi font for
Rockbox:

     http://www.rockbox.org/tracker/9059

>> The font can easily be built
>> from the original .hex files without those fillers though; leave out
>> blanks.hex when concatenating all the .hex files for the hex2bdf
>> script. In other words, do something like
>>
>> cat *.hex | sort | hex2bdf >unifont.bdf
>
> I think our problem is that we've been using the BDF from your site
> directly, rather than "compiling" our own from the .hex sources.
>
>> I have just put together a .hex file (which I can easily use to
>> generate a .bdf file) that shows each code point in the PUA as a
>> four-digit hexadecimal number in a box. That way if anyone does
>> encounter a special PUA character, at least they will know it is
>> there. You might want to use that instead, or just display the
>> substitution character.
>
> I think we'll opt for the substitution character, since including the
> hexadecimal number boxes will bloat the font size for little gain.
>
Okay, I'll leave out the 6400 PUA glyphs.

>> The combining diacritical marks aren't properly superimposed over a
>> preceding character in the BDF version. I did get them working
>> properly in the TrueType version.
>
>I don't even know how Rockbox' font handling deals with combining
>characters - if someone else could weigh in on that, that'll be great.
>
Combining characters are a little tricky. Unifont is dual-width; glyphs
are either 8 pixels wide or 16 pixels wide. In addition, most combining
characters are 8 pixels wide but some are 16 pixels wide (for example, a
breve mark that is supposed to appear above "oo"). There's a list of
all the combining characters in Unicode 5.1's Basic Multilingual Plane
in the source tarballs on my site in font/ttfsrc/combining.txt. There
are 975 such characters in the BMP. Feel free to use my combining.txt
list with Rockbox. For a more complete solution that tries to handle
multiple widths, you can look at this code from Markus Kuhn:

     http://www.cl.cam.ac.uk/~mgk25/ucs/wcwidth.c

>> If your release is soon, I don't know if I will be able to fix that in
>> the BDF version in time.
>
> It's probably too late to make it for the upcoming release (which should
> be on the 23rd, if nothing goes wrong), but since we're currently on a
> 3-month release schedule, the next release isn't too far off.

If the release is that soon then it would be better to wait for the next
release so there will be time to test the new version of the font. I'll
put a version together soon for you to test.


Paul Hardy
Received on 2009-03-12