Re: [css3-writing-modes] vertical orientation and UTR50

On 07/03/2012 05:10 AM, John Daggett wrote:
>
> Koji Ishii wrote:
>
>>> I don't think there's a need for a separate HO property and have
>>> posted a message in the UTR50 forum stating this. [1]  In fact,
>>> there's no role for HO in defining the behavior of the
>>> 'text-orientation' property since this property only affects
>>> vertical runs, *not* horizontal runs.  So the sentence starting
>>> with "The one exception..." can be omitted entirely.  In vertical
>>> runs, Mongolian and Phags-pa are displayed upright, just as the
>>> MVO/SVO reflects.
>>
>> Well, you know far more on Mongolian than I do, so I'd like to trust
>> you, but other two I also trust -- Laurentiau and fantasai think
>> Mongolian and Phags-pa should be rendered rotated, so I hope you can
>> find a consensus in the forum. I guess it's just difference of
>> visual orientations and rendering orientations, maybe wrong, but
>> it's a UTC's issue.
>
> Koji, are you sure this is what they are thinking?  These are
> *vertical* text runs we're talking about in the context of
> text-orientation.

I'm not sure what the confusion is here, but Mongolian and Phags-pa
rotate between horizontal and vertical orientations. Their rendering
is very complex due to contextual shaping (more so than Arabic) but
is identical in horizontal and vertical modes aside from the 90deg
rotation of the text run.

It is my understanding from talking with Martin Heijdra that this
rotation is expected to be done by the text engine, not by glyph
substitution in the font. Therefore the computation that is actually
used needs to represent the orientation with respect to the horizontal,
not the orientation with respect to the Unicode code charts (which
show the text in vertical mode).

I think your argument is that Mongolian and Phags-pa fonts use vert
substitution, and therefore should be typeset upright. Right? It's
possible, but given how many problems you've pointed out with the
interaction of features in vertical writing, seems unlikely to be
something people would rely on. :) Regardless I would trust Martin
and Microsoft on this matter, since they are the ones with real
experience working with Mongolian. If they say to use vert, then
we'll say to use vert. But since they say to typeset sideways, we
are saying to typeset sideways. And that means we need to account
for HO.

>> HO was resolved on the last UTC conference, it may not survive as
>> you say, but we can remove from our spec if they were removed from
>> UTR#50. We were there too, and supported the resolution, so not
>> using HO looks strange to me. I was ok either if it was an
>> informative text, but if the text is normative, I think we should
>> follow UTC's resolution.
>
> Look at the data, no implementation needs to use that data, they will
> realize immediately that the data reduces to "if the script is
> Mongolian or Phags-pa handle the horizontal case differently".
>
> I'm completely at a loss to understand why you requested a separate
> property value when expressing that very simple condition is
> sufficient.

For the horizontal orientation, we said that either some prose in
the spec or a new property would be fine, as long as it was explicit
enough that the reader had to make no inferences as to which codepoints
were affected. The UTC opted for a property.

~fantasai

Received on Tuesday, 3 July 2012 19:04:03 UTC