Re: OBJECT, inheritance, and rendering

Braden N. McDaniel wrote:

>And how is this not, functionally, a default style sheet?

It does not follow that because a UA has default style values, every
document has a default stylesheet. Note "cascading" in the name. The UA has
default values which can vary from UA to UA, the user can modify these
values, and a document can modify them further. If we accept your semantics
it could be argued that the default stylesheet for the included document
should consist of the values at the point of inclusion.

>"Independent" is a strong word. How strongly do you mean it? Do you mean
>here that the background color should be the UA default, or that it should
>be "transparent", and the color specified in the host document should show
>through?

Default, or as specified by the user. The idea of a UA algorithmically
modifying foreground colors in the included document to provide contrast on
an unexpected background is unappealing to me.

>Given (1) the conventional behavior of transparency in inclusions, and (2),
>the fact that the initial value for "background-color" is "transparent",
the
>former interpretation strikes me as inconsistent with the established
>behavior for other media types.

From the CSS2 recommendation, section 9.1.1*: "There is at most one viewport
per canvas..." The default color of the viewport's canvas is determined by
the UA, and it is not conventionally transparent (when you declare HTML and
BODY background to be transparent, do you see your desktop shining
through?). A complete HTML document must be rendered within its own
viewport. When an OBJECT is an HTML document, the object becomes, in effect,
an IFRAME. Rendering of HTML documents in the IFRAME element corresponds to
what I'm suggesting is reasonable.

>And I submit that only the latter interpretation requires no change to the
>CSS spec. It seems to me that the former interpretation would require that
>the initial value for "background-color" be changed from "transparent" to
>"implementation defined".

Nope, only that the nested viewport have its own canvas, as specified in the
CSS2 Recommendation.

>As stated, per my interpretation of the specs, this is not entirely
>accurate. The host document *can* affect the background color of the
>included document, but not the foreground color.

I see the included HTML document in a nested viewport with its own canvas,
ala IFRAME. I believe this is, at the moment, the conventional behavior of
transparency for text/html inclusions.

David Perrell

* <http://www.w3.org/TR/REC-CSS2/visuren.html#viewport>

Received on Friday, 31 July 1998 22:11:34 UTC