- From: David Perrell <davidp@earthlink.net>
- Date: Wed, 16 Apr 1997 11:33:04 -0700
- To: "Douglas Rand" <drand@sgi.com>
- Cc: <www-style@w3.org>, "MegaZone" <megazone@livingston.com>
Douglas Rand wrote: > I think the entire problem here is the one Gavin alluded to. The > document structure and the rendering do not have to be one to one > related. For example, I definitely format floating tables the way > you've described, but the document structure still has a para, a table > (not contained in the para) and some following text. That's the > document structure. The rendering structure has a para with a floating > element pointer in it and the table is held separately. The two just > aren't related. And in fact I have a current "bug" because the floating > table forces a paragraph break. When you put the table inside the paragraph and float it, N and IE treat it the same as an IMG, except that it does not have intrinsic dimensions. So it appears that the table is then structurally a sibling of the paragraph. Which seems to me to be a reasonable relationship. > What we could do is make a statement that floating elements force the > user agent to act "as if" they hadn't been in the flow after all. We > could even codify where they float including the behavior where if > they're first on the line they align with the top of the line, and > otherwise with the bottom of the line. But it'll be a hack. Gross! David Perrell
Received on Wednesday, 16 April 1997 14:43:16 UTC