Hacker News Reader

Comment by akdor1154 | Hacker News Reader

Interested in tangent replies to this, even if such fonts are artistically unsuitable for games - are open source fonts OK for Japanese? I understand that Unicode denotes single characters for both Chinese and Japanese (and Korean outside of Hangul?) even though there are differences between how nations write these 'single' characters, so the result is a Unicode font will look like a Chinese font, or a Japanese font, but not both.

How do the big Unicode OSS fonts like Noto, Deja Vu deal with this?

View on HN

IMO: quality wise, FOSS fonts are fine(n=1) - it doesn't have to be a proprietary font but do have to be the right ones of CJK.

- In cases with the Noto project, they've given up long ago on literal singular font in both name and file to cover every languages; they have bunch of variants as different fonts(TC,SC,HK,JP,KR,...), available as both individual files as well as combined files[1].

- In cases with most Latin fonts like Deja Vu - I think this is where bulk of "wrong font" problem comes from. They don't support CJK at all, and operating systems often handle fallback to system defaults for missing characters, but sometimes OS falls back into wrong CJK fonts and cause text mIshmashlnG[2] problems or show bunch of square placeholders(aka tofu). The latter is usually solved by installing Noto fonts, and former by specifying what fonts are to be used for fallbacks(anyhow it's done, it can be OS or game engine dependent).

- There are also bunch of domestically made fonts such as IPA Gothic; they usually only support the targeted one of CJK + ASCII(+ISO-8859-1). Most Japanese-English _bi_ lingual contents take this route, which had lead to curiosity by some as to why Japanese content creators appear to have unique and specific taste for fonts[3] - in fact they're just reusing the font files already in the project folder.

- For creative purposes such as games, FOSS fonts sometimes don't cut it, but that is not unique to the Japanese language.

1: https://cdn-ak.f.st-hatena.com/images/fotolife/s/satob/20231... | https://archive.is/6qeVm

2: https://blog.enjo.life/wp-content/uploads/2022/08/fonts-and-... | https://archive.is/jSQzn

1: https://www.youtube.com/watch?v=8dwSydGAJsk

> I understand that Unicode denotes single characters for both Chinese and Japanese (and Korean outside of Hangul?) even though there are differences between how nations write these 'single' characters, so the result is a Unicode font will look like a Chinese font, or a Japanese font, but not both.

This is "Han Unification", a terrible idea from early in the development of Unicode. The idea was so bad that the affected glyphs are now also given always-Chinese and always-Japanese Unicode points, making it possible to, for example, compare a Chinese character to a Japanese character in the same document.

But the fix exists. You can specify that you have no idea what you're trying to write by coding it as U+76F4 (直). Or you can specify a Chinese character by coding U+FAA8 (直). Or you can specify a Japanese one by coding U+2F940 (直). There isn't actually a reason you'd want U+76F4 - it's just a dead, useless unicode point - but we can observe here that my default font doesn't include a glyph for either U+FAA8 or U+2F940 even though U+FAA8 is by definition identical to U+76FA (since this is a Chinese font).

>There isn't actually a reason you'd want U+76F4

The fact that it was the only one that properly rendered for me is an actual reason.

And just think, that will be true about half the time!

You don't want U+76F4, you want buggy Asian fonts to be fixed.

(On second thought... did it render correctly for you? Did you check?)

Every single time I reload the page the other characters are a broken rectangle, so experimentally it's 100% of the time I'll need that character. What I want is to not see a broken rectangle. One can't say that U+76F4 is useless when it clearly is serving a use for my computer.

OP means it's the only one that renders at all. It could be wrong but the other two are just missing-glyph boxes which are definitely wrong. Those code points appear unusable as a practical matter.

On default modern Android I get the same: Only U+76F4 renders, not the other two.

Every variant is a different font. IMO Han unification is one of the greatest failures of Unicode.

Noto Sans comes in SC, TC, HK, CJK JP and CJK KR variants. There's no Noto Serif for CJK characters, nor does Deja Vu support them.

> There's no Noto Serif for CJK characters

Yes there is: https://fonts.google.com/noto/specimen/Noto+Serif+JP

Reply on news.ycombinator.com