RE: Pixels 'n points (CSS1 browser test)

> Chris Lilley [SMTP:Chris.Lilley@sophia.inria.fr] wrote:
>On Jun 27, 11:18am, Chris Wilson (PSD) wrote:
>> Douglas Rand [SMTP:drand@sgi.com] wrote:
>> >builds the rendering structure checks for displayFLOAT for inline or
>> >block elements (my internal value for the property) and sticks the
>> >content in a floater container.  It's relatively simple and few
lines
>>
>> True; but we can't alter the document structure like that,
>
>It isn't altering the document structure. It isn't even  dynamically
>modifying the stylesheet.

It is not intended to alter the SGML document structure, but hey,
"sticks the content in a floater container" sure sounds like it's doing
something to the internal document structure to me.

>> because we
>> need to be able to persist.  (The HTML rendering engine in IE4 is
also
>> an authoring system - e.g. it's used by Outlook Express, the email
>> client in IE4.)
>
>I don't see how that would conflict with being able to statically
declare
>floated elements in a style sheet.

Sorry, my message was apparently a little muddled.  We cannot simply
stick in the easy floating container that almost everyone has
implemented (a single-celled table), because that would alter the
document structure, and without a fairly significant coding effort would
probably be indistinguishable from a real single-celled table, which
would cause us major problems when we go to persist the document.  I'm
not saying this is an external problem; but it is a significant
requirement on our implementation.  Consider this an apology for this
not being an easy problem for us if you wish.

>> >I think some properties actually *are* problematic,  for example
>> >vertical-align applied to textual objects in paragraph flows really
>> make
>> >only modest sense.  The definition of vertical-align also doesn't
>> >correspond to common practice,  started with Mosaic,  of carrying
the
>> >top and bottom text  limits for the line as the line is formatted
from
>> >left to right.  Thus vertical-align can lead to circular
dependencies,
>> >which is not good, IMO.
>>
>> I agree whole-heartedly with you on this one.
>
>So, you would fix this how?

Sorry, I don't have the time to address this at great length right now;
try me again in a couple of weeks.  I guess my primary problem is that
there are a few things left unsaid about how vertical-align will affect
the parent's line box height.  Perhaps all that is needed to clear my
mind is the right set of diagrams - it's unclear to me on closer
examination whether there are circular references or not.

	-Chris
Chris Wilson
cwilso@microsoft.com
***

Received on Monday, 30 June 1997 11:47:11 UTC