Re: [css-overflow][css-break] pathological fragment box generation

On 4/6/13 9:36 AM, "L. David Baron" <dbaron@dbaron.org> wrote:

>On Thursday 2013-04-04 20:59 -0700, Alan Stearns wrote:
>> So the next step is to consider 0-height fragment boxes. Is there
>>anything
>> needed for that case if you are choosing to slice?
>
>I'm inclined to think that css3-break should specify behavior such
>that there is *never* a situation where the next fragment starts in
>the same situation as the current fragment, even if the fragment is
>zero height.

I agree that this should be part of the css3-break specification. I
proposed some text for this a few months back [1] which would allow
zero-height fragment containers (or those too small to fit anything) to be
skipped. Your idea of making progress at each fragment opportunity has
some clear advantages, but I find it strange to have a zero-height
fragment container consume anything, particularly if it causes the next
fragment to be missing some top margin or a slice of itself.

Since we already allow some differences in this area (the UA can choose to
slice a monolithic element or place the entire thing in a fragment
container that then overflows), perhaps we could allow differences here as
well? We could say that progress must be made in laying out the fragmented
flow over the fragmentation context, but leave it open as to whether this
progress needs to occur for each fragment?

>
>In other words, I would have a model where if the current position
>is at the top of a page/column/fragment (nothing placed yet), then
>something must be placed.
>
>If the (innermost) thing being placed is a a line (or portion
>thereof), a replaced block (or portion thereof), or margin, border,
>or padding, then placing at least one of these would be required.
>The height consumed for the blocks containing such object (whether
>fixed or auto height) would be determined by the height of the thing
>placed.
>
>On the other hand, if the thing that is placed is a portion of a
>fixed height that does not contain one of these things, such as the
>space below the lines in a fixed-height block that is taller than
>the space needed for its lines, then if that portion of height would
>be 0 (from the fragment height), then the line-height is used
>instead, and if that's 0 then the initial value of font-size is used
>instead.  I think my proposal is that these would be fallbacks
>rather than mimimums (that is, it would be possible to place less
>than these amounts if there were something at a specified but
>nonzero size).

Thanks,

Alan

[1] http://lists.w3.org/Archives/Public/www-style/2012Sep/0304.html

Received on Saturday, 6 April 2013 21:15:29 UTC