W3C home > Mailing lists > Public > www-style@w3.org > September 2015

Re: [css-writing-modes] computed value for text-orientation: sideways or sideways-right

From: Florian Rivoal <florian@rivoal.net>
Date: Sat, 19 Sep 2015 14:15:06 +0900
Cc: 馬場孝夫 <baba@bpsinc.jp>, Koji Ishii <kojiishi@gmail.com>, Gérard Talbot <www-style@gtalbot.org>, Jonathan Kew <jfkthame@gmail.com>, W3C www-style mailing list <www-style@w3.org>
Message-Id: <D8D3FFB9-E6C4-457E-9423-000B87CBC9B0@rivoal.net>
To: fantasai <fantasai.lists@inkedblade.net>

> On 19 Sep 2015, at 10:44, fantasai <fantasai.lists@inkedblade.net> wrote:
> 
> On 09/18/2015 09:31 PM, 馬場孝夫 wrote:
>> 
>> However, personally I can not really agree to those new writing-mode values.
>> 
>> Everything which is become possible by 'writing-mode: sideways-lr/rl' is also
>> able by 'writing-mode: vertical-lr/rl' + 'text-orientation:
>> sideways-right/left'.
>> 
>> If we reintroduce 'text-orientation: sideways-left' in future, those new
>> writing-mode values are no longer needed.
>> 
>> In my opinion 'writing-mode: sideways-lr' is not really needed just now
>> because it can be roughly replaced with 'transform: rotate'; therefore,
>> 'writing-mode: sideways-lr/rl' should be dropped.
> 
> CSS is a world-wide technology. Therefore it needs to accommodate
> the typesetting requirements of the whole world, not just of Japan
> and China. Vertical text is used in many other writing systems:
> it is commonly used for captions on the side of figures and tables,
> and for table column headings.
> 
> Transforming text does not work because it does not allow for
> correct layout of the text, such that the element is sized to
> fit its text and the surrounding content makes exactly enough
> room to fit the element.
> 
> Transforming text also does not work because certain effects
> that rely on a true coordinate system will not work correctly.
> For example, OS-styled buttons will be shadowed incorrectly
> if rotated 90deg.

I believe the suggestion was not to use 90deg rotate from writing-mode:horizontal instead of writing-mode: sideways-lr, but to use 180 deg rotation from writing-mode:vertical-rl;text-orientation: sideways. 

Even then, your point about elements relying on the true coordinate system remain valid.

I am also not sure that fragmentation would be identical between a rotated writing-mode:vertical-rl;text-orientation:sideways; and and actual writing-mode: sideways-lr. Interaction with floats would also be different, and probably other things I'm not thinking off of the top of my head.

Also, from a usability/learnability point of view, 

  writing-mode: vertical-rl; text-orientation: sideways; transform: rotate(180deg);

is much worse than

  writing-mode: sideways-lr;

Anyway, it is a nice hack that comes fairly close, but not a full substitute for the real thing.

> I know you are anxious for the features that Japan requires to
> be completed. However, this working group is not working only
> for Japan, and we must reasonably accommodate the rest of the
> world as well.
> 
> I also would not worry too much about these new keywords--
> compared to the rest of the work of making vertical text happen,
> they are not so difficult--particularly sideways-rl, which has
> the same effect as 'writing-mode: vertical-rl; text-orientation:
> sideways'.

I would even make a stronger statement than that: sideways-left is one
of the things delaying the progress due to difficulty of implementation,
and the complexity is due caused by opening possibilities (changing to
sideways-left inline) that are not justified by use cases.

Since indeed need to support the non CJK use cases well, the choice is not
between having the sideways writing modes or nothing, but between the
sideways writing modes or the sideways-left text orientation.

Since the sideways writing modes are much easier to implement, making
this choice is also good for Japan, since it means writing modes will
be finished earlier.

>> Do you mean that the values should be arranged like below?
>> 
>> - 'sideways-right': [dropped]
>> - 'sideways-left': [dropped]
>> - 'sideways': rotate clockwise
>> - new value such as 'sideways-reverse': rotate counter-clockwise [future level]
> 
> Yes, I believe this is what was meant.

It was.

>> If this my understanding is true, it seems to be consistent.
>> However, since I think auto-switching mechanism is still needed as I wrote in
>> above, I feel the original keywords look better.
> 
> I understand the desire for symmetry, and I'm not opposed.
> 
> I think I would like to investigate the backwards-compat impact
> of switching to 'text-orientation: sideways-rl' and dropping
> the other keywords, since that would be a little easier to type,
> but also clearer and more consistent with the corresponding
> writing-mode' keywords... what is your opinion on that, if it
> is feasible?

I am not strongly opposed either way. However, I think than
symmetry between sideways-left and sideways-right is overstated.
sideways right just affects glyph orientation, while sideways-left
also affects the baseline orientation and line progression direction.

So I have a preference for something like sideways and sideways-reverse
over -left and -right.


 - Florian
Received on Saturday, 19 September 2015 05:15:39 UTC

This archive was generated by hypermail 2.4.0 : Friday, 25 March 2022 10:08:57 UTC