Re: [css3-flexbox] getting multiline flexbox back into the spec

On 7/06/2011 11:04 PM, Eric A. Meyer wrote:
> At 13:56 +1000 6/7/11, Alan Gresley wrote:
>> On 7/06/2011 5:36 AM, Eric A. Meyer wrote:
>>> At 19:22 +0000 6/6/11, Alex Mogilevsky wrote:
>>>> Specifying direction of flexbox has nothing to do with specifying
>>>> direction of text flow within these boxes, therefore text direction
>>>> properties should not be used for that purpose. Flexbox should have
>>>> its own way to specify its layout direction.
>>> I agree with this 110%.
>> Eric, do you want another DOCTYPE switch scenario to surface in the
>> next decade due to incorrect concept of direction?
> You lost me there. I don't see how having layout order be potentially
> independent of writing direction triggers a switch scenario.

That's the thing, you said layout order where Alex is saying layout 
direction. The use of the word 'direction' complicates things.

> Has
> absolute positioning raised that possibility and I missed it?

Are you talking about CSS3 positioning? Currently it's not RTL or 
vertical writing mode friendly.

> Will grid
> layout similarly do so? Will regions?

Not sure how to answer this since I am now lost. I do know that Grid 
layout is not CJK dual layout friendly since an element that has columns 
establishes a BFC.

>> Base direction is the master of direction. Flexbox should not have it
>> own concept to specify its layout direction.
> Why not? As an example, 'box-direction: reverse' (to draw an example
> from the 2009 spec) defines a layout direction different than the
> standard layout direction. I've used it and it doesn't threaten the
> writing direction. It just reverses the layout of the boxes from what
> would be the default.

I OK with that. My main concern is this. Does RTL (via the attribute 
dir="rtl") change the ordering of the reverse layout? Does everything 
work in reserve in RTL?

> Similarly, 'box-ordinal-group' lets you throw boxes hither and thither
> within the overall flexbox, completely independent of writing or other
> layout direction, and I regard that as a good. It's useful for, say,
> placing a "sticky" post at the top of a forum while preserving
> chronological ordering in the source, or for switching around column order.

Agreed. I like the idea of sticky parts.

> I would find it deeply odd to have to use a writing-direction property
> to define layout order. That would be like having to use 'text-align' to
> center element boxes.

Agreed. I just wondering when a spec will solve the problem which I 
initially mentioned on this list [1] in 2010 and where Koji said that 
the multi-column could solve this. I can't test because IE10 has odd 
behavior when combing vertical mode and multi-column (blocks jump around 
crazy, nor knowing if the should be vertical or horizontal). I just 
wondering if flexbox could solve it.


Alan Gresley

Received on Tuesday, 7 June 2011 14:30:46 UTC