Re: CSS: %% length unit. Proposal. Some clarifications.

> > Right?
> 
> No, with the table layout algorithm (as implemented by UAs) width:100%
> means "make this column as wide as possible given the constraints of the
> other columns".

Letter 's' in your 'UAs' is just mistyping I guess. Only Mozilla is trying to render them properly (from Mozilla point of of view).  
Yep, 'cause "given the constraints of the other columns" has strong smell of fuzzy logic.

> 
> This is actually very poorly described by CSS2 right now, largely because
> nobody actually understands the table layout algorithm, which makes it one
> of the worst parts of CSS. Most implementations are developed by reverse
> engineering the more popular implementations.

In my opinion this happened because  table*-styles are "artificial" entities. Brave attempt to close holes. They conflict with e.g. box model. They create too many "if", "then", "except" across the whole specification.

Lack of  %% natural "fluid" mechanism in CSS was the main motivation to introduce them I guess.

> It's certainly a lot better than you have explained in the past. What do
> you mean by "line" exactly? Are floats and other overlapping elements part
> of the "line"? How do you determine what is on a line? What if there is no
> line box?

Exactly as it is right now. "Line" means "line box", floats if exist constitute left/right margins of the line box.

> 
> What if the container width is less than the width of the word "from:"?
> 

They will wrap as words.

Follow INPUTS on the first line :
Width #1

(http://terrainformatica.com/w3/p2/width1.jpg)
Width #2

(http://terrainformatica.com/w3/p2/width2.jpg)
Width# 3


Floaters/ left side:

(http://terrainformatica.com/w3/p2/width4.jpg)

Floaters/ both sides:
(http://terrainformatica.com/w3/p2/width5.jpg)

Andrew Fedoniouk.
http://terrainformatica.com