W3C home > Mailing lists > Public > www-archive@w3.org > December 2019

Fwd: UAX#50 conformance: Is it possible to update existing fonts without causing damage to existing non-CSS applications?

From: MURATA Makoto <eb2m-mrt@asahi-net.or.jp>
Date: Thu, 12 Dec 2019 13:47:52 +0900
Message-ID: <CALvn5EBupPoNOB9NvYTNzYXwi4nRvHQMNiyks-ZZzdEU6po+sw@mail.gmail.com>
To: www-archive@w3.org
---------- Forwarded message ---------
From: Taro Yamamoto <tyamamot@adobe.com>
Date: 2019年12月2日(月) 17:04
Subject: RE: UAX#50 conformance: Is it possible to update existing fonts
without causing damage to existing non-CSS applications?
To: MURATA Makoto <eb2m-mrt@asahi-net.or.jp>, Nat McCully <
Cc: Florian Rivoal <florian@rivoal.net>, fantasai <fantasai@inkedblade.net>


Nat might have different ideas, but I would like to make some comments.

> It appears that some fonts specify the vert feature in a
way inconsistent with CSS WM and UAX#50.  It becomes
impossible to use the glyph as specified in the Unicode
chart, since the vert feature always provide a different
glyph.  I learned this from Murakami-san.

In Japan, proprietary digital fonts for Japanese have been used since the
late 1960s. Japanese fonts for the PostScript interpreter appeared in 1989.
There have been Adobe's Type 1-based Original Composite Fonts, naked
CIDFonts, sfnt-wrapped CIDFonts on PostScript interpreters and the
Macintosh, and TrueType fonts and OpenType fonts on PCs, the Macintosh, and
some other operating systems. Today's fonts are based on the long history.
Also, there can be vendor-specific or typeface-dependent elements in
today's fonts, and many of them had been decided, before UTR #50 was
written. This historical nature applies also to how layout software handles
vertical postures.

As I have repeated, the switching of glyph postures in Japanese typesetting
are ruled mainly by application software. The area that the 'vert' feature
really can control is relatively small.

So, what you wrote in the "Document" is only half the domain of the
possible problems of vertical glyph posture inconsistency with UTR #50.
For example, If you type a COPYRIGHT SIGN U+00A9 character in a vertical
line on some page layout application software, the result posture will be
often  different from that defined by UTR #50. These applications treat it
as an R-vertical-posture character, and rotate it 90 degrees, but the
character must have the vertical posture of U according to UTR #50. This is
only ONE example of the inconsistency problems that you are concerned
about, but the 'vert' feature in font is NOT responsible for this problem
that occurs to the U+00A9 character at all, because the character is NOT
included in the 'vert' feature. This problem of U+00A9 is not a 'vert'
problem at all.

Still, I know everyone cannot deny that as you wrote, if the 'vert' feature
of a font instructs the vertical posture of an upright glyph in the font
mapped to a character whose vertical posture is defined to be U or Tu by
UTR #50, to be changed and to produce a rotated version of the glyph
instead, it is clear the result will be incompatible with the definition of
UTR #50, I agree that this is an incompatibility problem caused by the
'vert' feature.

(On the other hand, this also means that as long as the 'vert' produces a
variant glyph shape whose vertical posture is still Upright, there will be
no problem, even if the vertical glyph design has some difference from the
original Upright shape with the effect of the 'vert' feature, usually the
vertical glyph shape is what is expected by the type designer. This kind of
glyph substitution by the 'vert' feature must be allowed.)

In your 'Document', some 'vert' related problems are reported, such as
U+2016, etc.

Although I do not yet check every glyph listed in the "Document", some of
them seem to have been traditionally included in the 'vert' table in
Adobe's and other Japanese font vendors' fonts, and such fonts may be
producing the incompatibility problems with UTR #50 that you have
described. In the sense, the 'vert' table needs to be modified in some way
toward the future, but as mentioned above, updating the 'vert' table won't
solve every incompatibility problems with UTR #50. Furthermore, any changes
will cause backward compatibility problems which can damage the
reproducibility of legacy documents and printed matters.

This is my current, personal, view on the issue. Nat may have some
different views.



Received on Thursday, 12 December 2019 04:48:34 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:36:17 UTC