RE: [css3-writing-modes] a third option for implementing logical properties

I agree with John and David that this isn't an easy one and we should be very careful to design this.

I understand John is discussing two things: one about basic idea of logical directions, and the other about logical properties as one of the implementation method of logical directions.

The basic idea of logical directions isn't new, and I don't think it's as difficult for authors as John said. It is difficult for us unfortunately, because it swaps axis, but the painting context (like images) doesn't change. Axis and painting context were not distinguished until logical directions came in, but now axis needs to be abstracted. I understand it is difficult to spec and to implement.

But for authors who create vertical or RTL text, the idea is helpful rather than confusing. One of the e-book standard, IEC 62448, uses logical directions everywhere, while the painting context is absolute. There are a lot of e-books already published in this format. MS Word also uses the same idea. Almost every word processor I've ever seen in Asia use the same idea. It's a proven idea for a long period, and I think these implementations of logical directions help authors a lot until today.

The difficulty for CSS comes from where we have already decided to use physical directions in some places, so we can't use the similar method as IEC 62448. We need to invent a new method to use both physical and logical directions appropriately, and I agree that this is not an easy task.

For your second argument--about current spec of logical properties--authors who don't need to mix RTL nor switch his text to vertical don't need to use logical properties, so nothing changed for them. Logical properties are still good to make their styles logical and more re-usable, but authors who don't want to switch to the model don't have to.

Authors who care about directions have to make additional tasks as you said. And you're right that they have to write both margin-left and margin-start for browsers that do not support logical properties. That always happens during the transition period to introduce a new technology. I remember I wrapped my styles with <!-- and --> for old browsers, but today I don't. Shouldn't we introduce CSS because it requires <!-- and --> for backward compatibility? Of course not. It doesn't deny the great value of CSS, and in the long run, transition period ends.

As you suggested, writing separate styles for each directions can be also an option. But I see a couple of issues here:
1. If we consider vertical and RTL mixed text, we have to write 4 stylesheets. Add Mongolian to 8. If we're adding one, the new one should be logical and cover the rests.
2. Although both solutions may require similar amount of work during the transition period as you said, this solution requires the same amount of work even after 10 years, while logical properties does not.

You're also right that we still have some technical issues unresolved, like the one David pointed out. They need to be addressed before the spec goes CR. But those issues don't deny the value of logical directions.


Regards,
Koji

Received on Friday, 22 October 2010 06:01:30 UTC