Good evening,
I'm trying to migrate a software of my employer from ISO-8859-1 to UTF-8. The software itself has some kind of UTF-8 support, but we didn't write the the code ourself, which seems me to bring where I am now (bah!).
I'm looking for a Unicode-friendly TTF font in Fedora and/or EPEL. At the moment, we are using "Sazanami Mincho", but users from Japan told us, that some of the characters would be represented wrong for Japanese, but correct for Chinese and/or Taiwanese.
In the end, we need replacements for something like Helvetica (sans-serif), Times Roman (serif) and Courier (fixed) - with as many characters supported as possible. It needs to be a *.ttf file, because we have to use ttf2afm(1) for getting the font into our software.
Beside of that, that font(s) should somehow cover (if possible?) the above mentioned issue with the same character in JP/CN/TW but with a slightly different representation depending on the country. Is there a font which is covering that? Is there maybe one (!) font which is somehow acceptable for all countries, meaning Japanese, Chinese and Taiwanese people - without any claims by users that some characters are not perfect? "Sazanami Mincho" has many characters (from what we got), but we've claiming users :(
If it's somehow relevant, because I don't understand JP/CN/TW, the software is a Enterprise Resource Planning system, that means it's used at business, at organisations etc. If JP/CN/TW has any characters which are not relevant to business, the font hasn't necessarily to support them then... ;-)
Do you have any suggestions or recommendations for a font that could fit our needs? Hope to hear something else than "design your own font" by you. Maybe we (Fedora) have native speakers for JP/CN/TW, who could help here?
Greetings, Robert
On Wed, 20 Oct 2010 23:57:27 +0200, "RS" == Robert Scheck robert@fedoraproject.org wrote:
RS> I'm looking for a Unicode-friendly TTF font in Fedora and/or EPEL. At the RS> moment, we are using "Sazanami Mincho", but users from Japan told us, that RS> some of the characters would be represented wrong for Japanese, but correct RS> for Chinese and/or Taiwanese.
Not really. it was based on the machine-generated and somewhat improved by the hand. but is a poor quality right.
RS> In the end, we need replacements for something like Helvetica (sans-serif), RS> Times Roman (serif) and Courier (fixed) - with as many characters supported RS> as possible. It needs to be a *.ttf file, because we have to use ttf2afm(1) RS> for getting the font into our software.
RS> Beside of that, that font(s) should somehow cover (if possible?) the above RS> mentioned issue with the same character in JP/CN/TW but with a slightly RS> different representation depending on the country. Is there a font which is RS> covering that? Is there maybe one (!) font which is somehow acceptable for RS> all countries, meaning Japanese, Chinese and Taiwanese people - without any RS> claims by users that some characters are not perfect? "Sazanami Mincho" has RS> many characters (from what we got), but we've claiming users :(
I'd personally say no because they prefer different typefaces. though Google Droid fonts may be a candidate for unified fonts. but I don't like it really. at least it's a bit far from the modern typefaces for Japanese. FWIW this efforts are came from Android project though, recently another better Japanese fonts are available there. this would means how it wasn't accepted, but anyway.
RS> If it's somehow relevant, because I don't understand JP/CN/TW, the software RS> is a Enterprise Resource Planning system, that means it's used at business, RS> at organisations etc. If JP/CN/TW has any characters which are not relevant RS> to business, the font hasn't necessarily to support them then... ;-)
RS> Do you have any suggestions or recommendations for a font that could fit RS> our needs? Hope to hear something else than "design your own font" by you. RS> Maybe we (Fedora) have native speakers for JP/CN/TW, who could help here?
I guess it would be better not think of having only one font for all of the languages. you could pull an idea from current default fonts in Fedora. those should be comprehensively better fonts so far.
RS> Greetings, RS> Robert
HTH, -- Akira TAGOH
Hello Akira,
thank you very much for your reply.
On Thu, 21 Oct 2010, Akira TAGOH wrote:
I guess it would be better not think of having only one font for all of the languages. you could pull an idea from current default fonts in Fedora. those should be comprehensively better fonts so far.
From what I got when searching with Google, only JP/CN/TW/KR are somehow special about the characters. For us, that would mean 4 fonts plus one for the rest of the world could do the job, right? Or did I overlook something?
Do you have a recommendation for a good font that we could use for JP, a good one we could use for CN, one for TW and one for KR? If we can't have one font for all, let's try to have one font per country - does this work?
The goal is to make our users in APAC stop complaining by using the best font out there. If that means, we need multiple fonts, it isn't that easy as we hoped and assumed, but then it is as it is.
Any font name recommendations?
Greetings, Robert
On Thu, 21 Oct 2010 09:18:04 +0200, "RS" == Robert Scheck robert@fedoraproject.org wrote:
RS> Do you have a recommendation for a good font that we could use for JP, a RS> good one we could use for CN, one for TW and one for KR? If we can't have RS> one font for all, let's try to have one font per country - does this work?
RS> The goal is to make our users in APAC stop complaining by using the best RS> font out there. If that means, we need multiple fonts, it isn't that easy RS> as we hoped and assumed, but then it is as it is.
RS> Any font name recommendations?
I presumed that you have your own Fedora box that has an answer for your question. anyway, just tried to pick up the font packages from comps:
For Chinese: <packagereq type="default">cjkuni-ukai-fonts</packagereq> <packagereq type="default">cjkuni-uming-fonts</packagereq> <packagereq type="default">wqy-zenhei-fonts</packagereq>
For Japanese: <packagereq type="default">vlgothic-fonts</packagereq> <packagereq type="default">vlgothic-p-fonts</packagereq>
For Korean: <packagereq type="default">un-core-batang-fonts</packagereq> <packagereq type="default">un-core-dinaru-fonts</packagereq> <packagereq type="default">un-core-dotum-fonts</packagereq> <packagereq type="default">un-core-graphic-fonts</packagereq> <packagereq type="default">un-core-gungseo-fonts</packagereq> <packagereq type="default">un-core-pilgi-fonts</packagereq>
whidh would be installed by default for supporting their languages. I suppose you don't necessarily need everything above. it's just there for preferences and differeent shapes and styles in some cases.
HTH, -- Akira TAGOH
On 10/20/2010 10:27 PM, Akira TAGOH wrote:
I guess it would be better not think of having only one font for all of the languages.
But Unicode has separate character ranges for Chinese, Japanese, and Korean now. Can't you use a font that has distinct glyphs for those characters?
Hello John,
On Thu, 21 Oct 2010, francis+fedora+fonts@thibault.org wrote:
But Unicode has separate character ranges for Chinese, Japanese, and Korean now. Can't you use a font that has distinct glyphs for those characters?
which font(s) in Fedora could provide that or would satisfy that? Again, whatever font or fonts we use, the characters/symbols in CN/JP/TW/KR should be as perfect as possible to avoid the user complaints, that we right now get when using "Sazanami Mincho".
Greetings, Robert
Le Mer 20 octobre 2010 23:57, Robert Scheck a écrit :
Good evening,
I'm trying to migrate a software of my employer from ISO-8859-1 to UTF-8. The software itself has some kind of UTF-8 support, but we didn't write the the code ourself, which seems me to bring where I am now (bah!).
I'm looking for a Unicode-friendly TTF font in Fedora and/or EPEL.
Unicode fonts do not exist (apart from some very specific experiments no one uses in real life). What exist is fonts that cover more-or less wide ranges of the Unicode standard (and remember the standard is revised regularly so those fonts need regular maintenance to follow standard changes). And those covered ranges typically overlap in less than convenient ways (it's not just font A + font B + font C, some ranges like latin get ridiculous over-coverage and others are only covered by a specific font, or are not covered at all).
So, to answer your question, you need to identify the unicode ranges that matter to you, and search for a font-set that provides the needed coverage. If you're lucky whoever did the UTF-8 conversion plugged your app in pango/harfbuzz/fontconfig, and fontconfig will to the font mixing for you. If you're un-lucky you're better hope all the unicode ranges you need are supported in a single font. Otherwise you'll have to either add fontconfig support yourself, or use different software instances with different hardcoded fonts for different regions.
CJK is a particular hairball as the same codepoint need to be represented differently depending on the region. Here locale+fontconfig support is a must.
Regards,