W3C home > Mailing lists > Public > www-style@w3.org > June 2012

[css3-writing-modes] vertical orientation and UTR50

From: John Daggett <jdaggett@mozilla.com>
Date: Thu, 28 Jun 2012 22:14:47 -0700 (PDT)
To: www-style list <www-style@w3.org>
Message-ID: <1996458942.20533550.1340946887716.JavaMail.root@mozilla.com>

The current draft of the CSS3 Writing Modes spec defines the values
for the 'text-orientation' property based on the proposed vertical
orientation properties in Unicode [1].  Specifically, 'mixed-right' is
defined to be based on the mixed vertical orientation property value
(MVO) and 'upright' is based on the stacked vertical orientation
property value (SVO).  This makes sense and I completely agree with

However, in section 5.1.1 [2] the spec defines these in terms of
variations on the proposed UTR50 versions of these properties, using
the terms MVOsimple and SVOsimple.  It links to a data file on
csswg.org which contains values that are different from the current
data file associated with Draft 5 of UTR50.  Elika says these are
edits that have already been generally agreed upon but have not yet been
published in a UTR50 draft.

I really, really, really don't like the idea of publishing our own
variation of UTR50.  Right now there's considerable discussion in
Japan via twitter etc of UTR50 [3] and the role of MVO/SVO in vertical
text layout.  I think some of this discussion may just be
misunderstanding about the intent and utility of UTR50 data but I
think part of it may be valid criticism that leads to
changes/restructuring of the data.  I think we should let that
discussion influence definition of UTR50 and not introduce more noise
into an already noisy discussion.  The proper place for discussion of
these details is in the Unicode forum for UTR50, not on www-style.

I realize that applications are being written that diverge in
behavior [4] and that some sort of convergence is necessary for
interoperability.  But given the current state of the discussion I
think the only way to draw matters to a conclusion is to wait for
those with opposing views of UTR50 data to resolve their differences. 
Publishing a CSS version of the data will not aid resolution.

It would be much better for this part of the spec to define the
behavior of 'mixed-right' and 'upright' given the specific values of
the MVO/SVO properties defined in UTR50 (i.e. R, T, Tr, Tu, U). The
spec contains the wording below but in an informative note that
doesn't fully explain how the logic is to be used:

> These two properties are derived properties using the
> pseudo-algorithm shown below:
>       if ((HO = "L") and (VO = "U")) then
>         VOsimple := "R"
>       else if (VO in ["U", "T", "Tu", "Tr"]) then
>         VOsimple := "U"
>       else if (VO = "R") then
>         VOsimple := "R"
> from draft #5 of [UTR50] with a few changes that were discussed with
> UTC.

It is simpler to describe what is meant in words rather than in
pseudo-code that reuses "R" and "U" to mean both "UTR property value"
and "resulting vertical orientation" and introduces terms like HO and
VO without defining them.

Proposed replacement:

  For the 'mixed-right' and 'upright' property values, vertical orientation is
  defined in terms of the corresponding Unicode property value, MVO
  and SVO respectively. If the orientation value is "R" then the glyph
  is rotated right.  If the orientation value is "U", "T", "Tu", or
  "Tr" then the glyph is displayed upright.  The one exception is for
  scripts like Mongolian for which special handling is required for
  the stacked case due to the vertical-only nature of the underlying

I think it might be better to omit reference to HO at this point, I'm
not sure that property is going to pass full Unicode committee
approval given that it's effectively defining a character property
that represents the condition [script == Mongolian or Phags-pa].

John Daggett
Mozilla Japan

[1] UTR50 Unicode Properties for Horizontal and Vertical Text Layout

[2]  Vertical Orientations in CSS3 Writing Modes

[3] Twitter discussion of UTR50 (in Japanese) - *lots* of posts!

[4] Screenshots of default rendering of Latin text within vertical Japanese text in EPUB apps
Received on Friday, 29 June 2012 05:15:15 UTC

This archive was generated by hypermail 2.3.1 : Monday, 2 May 2016 14:39:00 UTC