U+FF3E (fullwidth circumflex) and U+FF40 (fullwidth grave) are too high in vertical text

[GoogleCodeExporter]

They are too close to the character before them.

Compare the screenshots, between SimSun and Noto Sans CJK SC.

Original issue reported on code.google.com by roozbeh@google.com on 24 Oct 2014 at 2:06

Attachments:

• noto.png
• simsun.png

[GoogleCodeExporter]

BTW, the sequence is <FF3F, FF40, FF3F, FF3E, 9A5A, FF3E>

Original comment by roozbeh@google.com on 24 Oct 2014 at 2:06

[GoogleCodeExporter]

My first (and only) question is "who uses these particular characters in 
vertical text?" If memory serves, this particular issue was discussed during 
the development of Noto Sans CJK.

Original comment by ken.lu...@gmail.com on 24 Oct 2014 at 3:24

[GoogleCodeExporter]

Ken, I don't have that information, but we really care about this getting fixed.

Original comment by rooz...@gmail.com on 24 Oct 2014 at 3:45

[GoogleCodeExporter]

Why not?  An author might want to use it to express a garbage text from broken 
software in his novel.  Or a popular science text might want to use it to 
typeset a short formula with logical XOR operator, like A^B, vertically.

Original comment by taken...@google.com on 24 Oct 2014 at 3:57

[GoogleCodeExporter]

Okay, but the reason for my reply (#2) is because the glyphs for these 
characters are the same as what is included in Kozuka Gothic, which was 
released almost 15 years ago, and an issue with these particular characters has 
never been raised.

Anyway, if the issue is only about their placement in vertical writing, the 
best way to address this will be via a 'vmtx' table override, which can be used 
to change the vertical origin, in terms of its position along the Y-axis. I can 
either make the glyph centered in the em-box along the Y-axis, or simply to 
make sure that it is confined within the em-box. Please let me know your 
preference.

Original comment by ken.lu...@gmail.com on 24 Oct 2014 at 4:55

[GoogleCodeExporter]

I guess confinement within the em box is enough, but I'm no CJK expert. 
Takenaka-san, do you have a recommendation or preference?

Original comment by roozbeh@google.com on 24 Oct 2014 at 5:11

[GoogleCodeExporter]

The reason why I am suggesting the option of centering the glyphs in the em-box 
along the Y-axis is because the upper and lowercase Latin and digits in the 
same range are already being overridden in this way so that they are centered 
in the em-box along the Y-axis. The effect is somewhat subtle for the uppercase 
Latin and digits, but is rather striking for the lowercase Latin.

Original comment by ken.lu...@gmail.com on 24 Oct 2014 at 5:21

[GoogleCodeExporter]

> an issue with these particular characters has never been raised.

I can imagine that, because vertical text is been usually for printing on paper 
or in PDF.  So compatibility of metrics with other major fonts doesn't matter.  
But I think it does matter if you exchange vertical text e.g. in EPUB format 
between different viewers / platforms with different fonts.

As for fix, I think just confined within the em-box, like the SimSun sample 
above, should be fine.

Original comment by taken...@google.com on 24 Oct 2014 at 5:46

[GoogleCodeExporter]

Roger that.

Original comment by ken.lu...@gmail.com on 24 Oct 2014 at 9:10

[GoogleCodeExporter]

Original comment by xian...@google.com on 12 Apr 2015 at 11:11

• Changed state: Fixed