Re: [css-grid] step zero in 10.5

On 02/28/2016 05:32 PM, Mats Palmgren wrote:
> The recently added step zero in "10.5. Grid Item Placement Algorithm"
> says:
> "This effectively changes the order of their boxes in the box-tree,
> and how the rest of this algorithm deals with the generated flex items."
> https://drafts.csswg.org/css-grid/#auto-placement-algo
>
> I suppose you mean "grid items"?  I don't understand what the latter
> part about "generated grid items" is trying to say though.
> Please clarify.
>
> Also, why do you elaborate here about how to do the ordering at all?
> Couldn't you simply say that this algorithm works on boxes in
> "order-modified document order"? a term used elsewhere in Grid
> and linked to its definition in Flexbox:
> https://drafts.csswg.org/css-flexbox-1/#order-modified-document-order
>
> I would much prefer if you use already established terms and
> definitions to avoid ambiguity (i.e. is the added text here in fact
> the same as the "order-modified document order" or is there some
> subtle difference?)

It's the same; we forgot to update this step to use the term
"order-modified document order" when we first defined the term.
We should fix this in both specs.

Note: The exact same wording is used in Flexbox:
   https://www.w3.org/TR/css-flexbox-1/#box-manip

> There is also this (new) note:
> "Note: Absolutely-positioned children of a grid container do not
> participate in grid layout, but are reordered together with any grid
> item children."
>
> I think it's a mistake to mention abs.pos. boxes here.  This algorithm
> doesn't apply to abs.pos. boxes at all.   Chapter  "11.2. With a Grid
> Container as Parent" already covers this and it's clear:
> "An absolutely-positioned child of a grid container does not affect
> the placement of other items or the sizing of the grid. However, it
> does participate in the reordering step (see order), which has an
> effect on painting order."
> https://drafts.csswg.org/css-grid/#static-position
>
> I think the added "reordered together with any grid item children"
> part adds confusion because it suggests that grid items are affected
> by abs.pos. box order somehow (or vice versa) which isn't the case.
> These two types of boxes can be sorted orthogonally.

I'm not sure what "orthogonally" means here, maybe "independently"?

But no, they cannot be sorted independently: the abspos items
interleave with the in-flow grid items in the z-order. The note
is here to call this out...

However, it appears that no implementation does this, so we
should maybe remove it from the spec. ^_^ We've filed this
as a separate issue.

~fantasai and TJ

Received on Monday, 11 April 2016 22:51:00 UTC