Re: Proposition. Positioning content with guidelines

Another notable difference between his proposal (guidelines) and grid I 
think I grasped is the fact guidelines do not have "proven" existence in the 
sense they may be ignored (resulting in a cell fusion) if there's an element 
who breaks them. How (and if) that behavior can be added to grid is an open 
question, however. The main problem is that a "grid separator" can be broken 
by the last element of the grid, resulting in a full recomputation of the 
layout of the grid, since two columns now have been merged. Merging column 
is possible due to a "grid-cell-stacking: flow" equivalent which makes all 
element inside a cell child of some anonymous box: when the box can't 
display all columns horizontally, some are merged and stacked vertically (or 
in flow, in function of what you specify).

While this may seems complex, there are cases where it helps to keep 
consistent look&feel in environements when you have no control in the final 
width of your grid. To achieve the same result using a grid, you need a 
media query or a script that change the grid-template in function of the 
size of the grid.




-----Message d'origine----- 
From: Phil Cupp
Sent: Friday, February 24, 2012 9:33 PM
To: Alexander Shpack ; www-style
Subject: RE: Proposition. Positioning content with guidelines

> From: Alexander Shpack [mailto:shadowkin@gmail.com]
> In continuation to the topic 
> http://lists.w3.org/Archives/Public/www-style/2010Sep/0097.html
> This is my second attempt, I changed the guideline declaration, make more 
> examples.

Hi Alexander,

Before jumping to a new solution I think it would be helpful to articulate 
the problems you'd like to see solved.  Maybe some of them already have 
solutions or some could be solved with minor changes to existing specs and 
maybe some problems shouldn't be solved with CSS.  It's difficult for me to 
reverse engineer what those problems are by studying the beginnings of a new 
specification.

>From your discussion with Francois REMY it seems one of the perceived 
>problems may be that everything must be explicitly positioned on the grid, 
>and for some scenarios it might be simpler (or necessary) to infer 
>something about the proper position of a grid item based on its source 
>order or other CSS properties.

Does that sound right?  If so, then I think Francois has it right: the grid 
auto placement feature is meant to address these issues, but still needs 
some work.  Now is a great time to highlight what scenarios are important to 
solve with the auto placement algorithm.

What would really be helpful for me is a list of layouts that you'd like to 
achieve that you think can't be accomplished through some combination of our 
existing layout types, and why it's important to achieve those using CSS (in 
particular with the grid).

-Phil