Re: [css-position] sticky only along one axis

On Mar 18, 2014, at 4:22 PM, "Tab Atkins Jr." <jackalmage@gmail.com> wrote:
> 
> On Tue, Mar 18, 2014 at 4:17 PM, Brad Kemper <brad.kemper@gmail.com> wrote:
>>>> On Mar 17, 2014, at 5:44 PM, "Tab Atkins Jr." <jackalmage@gmail.com> wrote:
>>>> On Mon, Mar 17, 2014 at 2:46 PM, Brad Kemper <brad.kemper@gmail.com> wrote:
>>>> What makes it hard? Don't TDs already look to COLs for style resolution?
>>> 
>>> Only for four specific properties (border, background, width, and
>>> visibility), which are conceptually styling the *column itself*, not
>>> the cells in the column.
>> 
>> Those properties are no more so "column itself" than positioning could be.
> 
> Yes they are.  Columns, themselves, can have borders, backgrounds,
> widths, and a visibility;

That makes it sound like the those styles are being applied to a separate rectangle. They aren't; they are being passed to the TD. 

> these are all taken into account when the
> whole table is being assembled and painted.
> 
>> They are still applied to the TD, not some tall rectangle sitting behind it spanning all the rows. And if you can select all the cells in a row and style them, I don't see why you shouldn't be also able to select all the cells in a column and style them. Conceptually it is the same, and technically it doesn't sound all that more difficult.
> 
> It is.  This isn't an arbitrary choice.  For example, which cells are
> in a column is a result of styling, itself, as you can change the
> 'display' value for individual cells.  

So?

> Further, allowing arbitrary
> styling on columns to mean "style the cells in the column" means that
> cells are now inheriting from *two* parents, which requires
> adjudication of some sort.  

That isn't the way I described how it should be done. Selecting the column should be more like a selector shorthand for selecting its cells. The cascade provides the adjudication based on specificity and order, as always, not inheritance. I'm not suggesting that table cells should change how they inherit. 


> It's possible that more properties could
> be applied to the column itself and dealt with at the same time as
> borders/etc are, but it's definitely not something you can do for
> arbitrary styles.

I didn't say anything about arbitrary styles. I was talking about position (specifically sticky and relative, if I didn't make that clear yet), which is being improved in the current draft wrt tables. I'd just like it improved a little more.

Received on Tuesday, 18 March 2014 23:53:31 UTC