Re: [css3-text-layout] margin-before/after/start/end etc. and :ttb pseudo-classes

On Jun 6, 2010, at 10:16 PM, fantasai wrote:

> On 06/06/2010 06:32 PM, Brad Kemper wrote:
>> 
>> On Jun 6, 2010, at 5:20 PM, fantasai wrote:
>> 
>>> On 06/06/2010 04:28 PM, Håkon Wium Lie wrote:
>>>> Also sprach MURATA Makoto:
>>>> 
>>>>  >   So,
>>>>  >
>>>>  >   >     :lrt   horizontal writing is supported and @dir has been set to 'lrt'
>>>>  >   >     :rtl   horizontal writing is supported and @dir has been set to 'rtl'
>>>>  >   >     :ttb   vertial writing is supported and the initial value of
>>>>  >   >            'writing-mode' is 'tb-rl'
>>>>  >
>>>>  >   is different from what I wrote as "pseudo-selectors based on values specified in
>>>>  >   documents or stylesheets ".
>>>> 
>>>> Right.
>>>> 
>>>> Does this mean you could live with the definition, as quoted above?
>>> 
>>> The last one doesn't make any sense to me. The initial value of 'writing-mode'
>>> is 'lr-tb' by definition.
>> 
>> I took that to mean that the mechanism to change the default writing mode, which
>> lies somewhere outside the document and style sheet (as we have most recently
>> been talking about it), has changed the mode to 'tb-rl', as though that was the
>> initial value.
> 
> Sorry, you'll have to explain what exactly you mean by that.
> 
> The initial value is what I get if I, as an author, set 'writing-mode' to the
> keyword 'initial', or, if I , as an author, set 'writing-mode' to inherit on
> the root element. Are you saying that this value should be changed by the UA?

Yeah, that's my understanding of how the wind has been blowing the conversation recently. I asked early on how the UA was doing this change that represents the user's preference. If the UA is monkeying with the 'writing-mode' via a style sheet that overrides the author's, then that is a different situation, in which the author can do nothing about:

• directional list markers (like colored triangles that point to the text), 
• accordion control images that point in the direction the thing opens, 
• border-images in which there is some sort of crown and/or foot in the image, 
• background-images that work a certain way with the direction and/or alignment of the text, 
• column numbers, widths, and heights of text boxes and other elements that are designed to fit in a certain minimum screen size and read comfortably without 2-axis scrolling, 
• etc.

Switching the direction naively based on *-start *-end, etc. might work better than nothing for some designs, and for others might fall far short. An author taking the care to put those in, instead of -right -left, etc. might also want the rest of the design to look nice when the primary scrolling direction is horizontal. He might still want a horizontal header across the top, but with article text running in horizontal columns under that. He might also want background images that are more targeted for an Asian audience. By having two separate blocks of rules (or  two separate style sheets) the author can also have a better chance of guessing/testing what the final design will look like, even if his UA or authoring tool doesn't support vertical text, and he worries less about having missed something in a large style sheet (like shorthands that don't have the words "right", "left", "top", or "bottom" in them). 

There were several posts that lead me to believe that the UA is expected to change the writing mode regardless of what the document or style sheet says.

Received on Monday, 7 June 2010 15:24:50 UTC