Apple's FontBook, `hyperglot` and Linux's `fc-scan` detect that some open source fonts built with Glyphs do not speak Japanese. Can you figure out the cause?

Glyphs
1

Hello!

I have tested three open-source fonts coming with Han script and Japanese scripts – hiragana and katakana:

Sarasa Term J

Apple’s FontBook, hyperglot and Linux’s fc-scan detect that this font speaks Japanese, as these scripts as well.

M++ and Tetsubin Gothic

Apple’s FontBook and Linux’s fc-scan detect that these fonts have Han and Japanese scripts, but do not speak Japanese. The command hyperglot detects that these fonts do not have these scripts nor speak Japanese.

Observations

Apple’s FontBook comes with a language label. At the language label “Japanese”, Sarasa Term J is labelled, but M++ and Tetsubin Gothic are not labelled, despite having Han and Japanese scripts because it claims that they do not speak Japanese.

I believe that there is a bug in these fonts, or maybe Glyphs has this bug. I am not confident. Please, can you figure out the cause of this problem?

2

As fare as I know those tools test the available glyph set. So there seem to be some glyphs missing that are deemed required.
And there are the “unicode ranges” in the OS/2 table. Can you check if those are set differently in those fonts?

3

@GeorgSeifert, I have justed tested with:

  • Testubin Gothic:

    Testubin Gothic 1
    Testubin Gothic 2

  • Sarasa Term:

    Sarasa Term

It appears that both fonts have these enabled Unicode ranges, but I noticed that these fonts have different code page ranges.

I have just checked the completeness of «Japanese» at Glyphs.

  • Testubin:

    Testubin Gothic 3

In comparison with

  • Sarasa Term:

    Sarasa Term 2

I checked with M++ font:

M++ Regular

The kana script is complete, but Apple’s Font View and hyperglot detect that this font does not have kana script. In the case of the command hyperglot, it should look like:

Languages in Hiragana
=======================

Ainu, Japanese, Okinawan
4

I meant the unicode ranges and codepage settings in the final font file. You can check it with the Font Table Viewer or ttx.

5

@GeorgSeifert

I am not confident that it is what you meant, but here are the screenshots, and click the image to enlarge it in a new table.

I will provide you with a zipped file with three open-source fonts for you to analyse. As your forum limits the file size, I generated a temporary shareable link to download:

6

You can see that the M++ fonts doesn’t set some of the Japanese unicode ranges. That is either a fault of Glyphs (by calculating it wrong), the designer (by manually setting up wrong Unicode ranges or missing out some glyphs that are needed).

What exactly do you mean by that? For me all three showed “Japanese” in the list of supported languages in Font Book.

And hyperglot seems to actually analyze the glyph set and some missing glyphs will make it tell you that it doesn’t support a language/script.