Re: [css-regions] responsive and semantic use of named flows

On 1/25/14, 1:49 PM, "Håkon Wium Lie" <> wrote:

>Alan Stearns wrote:
> > >Wouldn't this be more naturally expressed by making the <aside> be
> > >kind of positioned float or exclusion, positioned below the fold, that
> > >the <article>s all flow around?
> > 
> > That depends on how you evaluate ‘natural’. As François points out, in
> > order to invent something new that would handle this particular case
> > floats, you would need to add a float feature that expressed this
> > 
> > ---
> > Float past the bottom of the view, unless the content doesn’t fit the
> > view. In that case, float to the bottom of the content.
> > ---
>This can be expressed with 'float-reference':
>  aside {
>    float: top;                   /* first, float me to the top */
>    float-offset: 100vh;          /* then, push me in the opposite
>direciton */
>    float-reference: multicol;    /* but not outside the bounds of the
>element */
>  }

>That is, by setting the float reference to 'multicol', you set the
>bounds of the float to be the element (and not, say, the page or column).

I think this is a fine sample to compare against, as far as I understand
the intent behind the properties. I think named flows and region chains
are better-described than page floats at this moment, but I am a bit

In my version, I set the height of the first fragment box with a single
max-height declaration, and use static positioning to achieve the effect.
The other three declarations set up the named flow and region chain.

In your version, you use three properties to set the position of a page
float. I’m assuming the ‘multicol’ value does not require the parent
element to actually be multicol.

In order to use my version, you have to understand named flows and region
chains. Once you do, you can use fragmentation to get other positioning
effects using all of the positioning schemes we devise for CSS.

In order to use your version, you have to understand how page floats work.
If a new positioning requirement crops up, more properties and values may
need to be added. And if (for example) you wanted apply the flexbox
algorithm to position a float, you’re out of luck.

Håkon - you and I certainly have our preferences, and I don’t think either
of will convince the other. I’d be interested to hear other people’s
opinions on this particular comparison.



Received on Saturday, 25 January 2014 22:27:28 UTC