Re: [css3-text-layout] New editor's draft - margin-before/after/start/end etc.

> But I also noted a smart proposal that fantasai wrote. See:
> http://lists.w3.org/Archives/Public/www-style/2006Jun/0084.html

Thank you for reminding us of the old proposal by fantasai.  
I like it and believe that it is completely free from implementation 
difficulties pointed out so far.  In particular, cascading does not become 
complicated at all. 

Her key idea is to separate the exisiting physical properties and 
the new logical properties as much as possible.  There are almost 
no interactions between the two groups of properties.  Then, 
the direction-model property chooses which group is in effect.

I think that this proposal is attractive for two reasons.

First, it does not make implementations difficult.  We only need a
new program for handling new logical properties.  That
program does not have to consider old physical properties.  The 
existing program for handling phsyical properties does not 
have to be changed.  We also need a program for computing 
the direction-model property and choosing which group of 
properties is finally used for displaying the content.

Second, it allows fallback by *existing* implementations.  
Stylesheet authors only have to write old physical 
properties for existing implementations (assuming 
horizontal writing) and new logical properties for new 
implementations (assuming whichever direction might 
be eventually chosen by the browser or user). 

> I am still thinking that DDAs is a good idea and Mozilla and WebKit's
> *-start/end implementations seem good enough.

I think that we should concenrate on fantasai's proposal.

> >  > You can't resolve before cascading, can you?  Or rather, you can't 
> >  > resolve before you have cascaded and maybe-inherited writing-mode.
> > 
> > Yes, 'writing-mode' must be processed to specified value before the
> > DDAs can be cascaded. That's an unhealthy inter-dependency between
> > properties, it seems.

fantasai's proposal is free from such unhealthy dependencies, as 
correctly pointed out by Murakami-san.

Cheers,
Makoto

Received on Sunday, 6 June 2010 02:19:35 UTC