Re: [CSS2.1] Yet another white-space issue with anonymous table objects

Sent from my iPhone

On Mar 9, 2009, at 5:52 PM, Boris Zbarsky <bzbarsky@MIT.EDU> wrote:

> Brad Kemper wrote:
>> I can imagine a reason: because the HTML source is in proper  
>> semantic order, but not in ideal order for the design, or there are  
>> many alternate style sheets for the same general HTML, and you want  
>> to move the middle block, perhaps even move it off screen, or have  
>> it appear to the right as a sort of tool tip when you hover over  
>> the body.
>> I think it is a reasonable expection that the two table cells act  
>> like siblings and that a div stuck between them does not act as a  
>> defacto "new row" marker.
>
> For what it's worth, that's not the behavior in any of Gecko,  
> Webkit, or Presto.  I won't comment on whether it's what the spec  
> expects, since my point was that the spec is completely unclear on  
> the matter.  I _can_ tell you that implementing the behavior you  
> suggest would be quite a bit of a pain in Gecko...  I can't speak to  
> other CSS implementations.

Actually, if I wrap those three DIVs in another one that has 'display:  
table row;' then I get what I described in Safari 3, which is plenty  
good enough with me, and I can see that without that it is two tables,  
not just two rows.

If I  then put 'left: 3em;' into B (with the table-row around all  
three), I get this in Safari 3:

AC  B

But in Firefox 3 it moves down a line (under the table), like this:

AC
        B

I don't have IE8 here at the moment, so I don't know what it does. I  
think what Safari does is more useful, but I don't know if it is the  
"right" thing.

> In any case, the concept of "sibling" as used in section 17.2.1  
> really needs a rigorous definition.

No argument from me on that point. Consistency between UAs would be  
good.

Received on Tuesday, 10 March 2009 01:33:29 UTC